AI - Surpevised Regression

Wind Power Forecasting

Importing

Exploratory Data Analysis

Preprocessing

Fitting Models

Results

data scientist | engineer

caio sóter

Wind power forecasting

Wind energy is the energy of the wind transformed into useful energy through wind turbines. This renewable energy source is widely used because it is an alternative to fossil energy, it is clean, does not produce greenhouse gases and can generally be used in various locations, however there are still some environmental and social problems attached, such as soil compaction and the noise emitted by the blades. In addition, wind energy suffers a lot from the fluctuation of winds and, therefore, doors are opened for the application of Machine Learning models to be used to make generation forecasts. Finally, this project aims to forecast the wind power generation of a wind turbine located in Germany with historical data from 2011 to the end of 2021.

Dictionary (Column)

• dt: Time series with timestep of 15 minutes.
• MW: Wind power (MW).

References

• kaggle dataset
• https://towardsdatascience.com/illustrated-guide-to-lstms-and-gru-s-a-step-by-step-explanation-44e9eb85bf21

1) Importing Libraries and Data loading

# Importing Libraries:
import pandas as pd
import numpy as np

import plotly.express as px
import plotly as pl
import matplotlib.pyplot as plt
import seaborn as sns

from sklearn.feature_selection import r_regression
from sklearn.preprocessing import MinMaxScaler, StandardScaler, RobustScaler
from sklearn.ensemble import RandomForestRegressor
from xgboost import XGBRegressor

import tensorflow as tf
from keras.layers import LSTM
from sklearn.metrics import mean_squared_error
from tensorflow.keras.layers import Conv1D,  MaxPooling1D

import os
from datetime import datetime as dt

# Loading dataset
df = pd.read_csv("data.csv")

# Copy of the dataset:
dado_horarios = df.copy()

2) Exploratory Data Analysis

# Let's see the first five rows:
dado_horarios.head()

	dt	MW
0	2011-01-01 00:00:00	3416.0
1	2011-01-01 00:15:00	4755.0
2	2011-01-01 00:30:00	4939.0
3	2011-01-01 00:45:00	4939.0
4	2011-01-01 01:00:00	4998.0

# Some information about data type and memory:
dado_horarios.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 385566 entries, 0 to 385565
Data columns (total 2 columns):
 #   Column  Non-Null Count   Dtype  
---  ------  --------------   -----  
 0   dt      385566 non-null  object 
 1   MW      385566 non-null  float64
dtypes: float64(1), object(1)
memory usage: 5.9+ MB

# Changing the type of the Time series column:
dado_horarios['dt'] = pd.to_datetime(dado_horarios['dt'])

# Dt column as index:
dado_horarios.set_index('dt', inplace=True)

# Some decriptive statistics:
dado_horarios.describe()

	MW
count	385566.000000
mean	3183.560481
std	3045.941591
min	0.000000
25%	905.000000
50%	2164.000000
75%	4529.877500
max	16727.500000

# Missing data:
dado_horarios.isna().sum()

MW    0
dtype: int64

As we can see above, there is no missing data.

# Aggregating information:
dado_horarios['month'] = dado_horarios.index.month
dado_horarios['year'] = dado_horarios.index.year

group = dado_horarios.groupby(["year", "month"]).mean().reset_index()

# Ploting the monthly average Wind Power by year:
fig = px.line(data_frame=group, x='month', y='MW', color='year', title="Monthly Average Wind power generation by year")
fig.show("png", width=1200, height=500)

Observations:

• Compared to the average energy produced in 2011, the energy produced in 2021 is much bigger for every month.

# Histograma da série temporal:
fig, ax = plt.subplots(ncols=4, nrows=3, sharex=False, sharey=False, figsize=(25, 20))

sns.histplot(x=group[group['year'] == 2011]['MW'], kde=True, ax=ax[0, 0])
sns.histplot(x=group[group['year'] == 2012]['MW'], kde=True, ax=ax[0, 1])
sns.histplot(x=group[group['year'] == 2013]['MW'], kde=True, ax=ax[0, 2])
sns.histplot(x=group[group['year'] == 2014]['MW'], kde=True, ax=ax[0, 3])
sns.histplot(x=group[group['year'] == 2015]['MW'], kde=True, ax=ax[1, 0])
sns.histplot(x=group[group['year'] == 2016]['MW'], kde=True, ax=ax[1, 1])
sns.histplot(x=group[group['year'] == 2017]['MW'], kde=True, ax=ax[1, 2])
sns.histplot(x=group[group['year'] == 2018]['MW'], kde=True, ax=ax[1, 3])
sns.histplot(x=group[group['year'] == 2019]['MW'], kde=True, ax=ax[2, 0])
sns.histplot(x=group[group['year'] == 2020]['MW'], kde=True, ax=ax[2, 1])
sns.histplot(x=group[group['year'] == 2021]['MW'], kde=True, ax=ax[2, 2])
ax[2, 3].set_visible(False)


ax[0, 0].set_title("Monthly average Wind power distribution of 2011")
ax[0, 1].set_title("Monthly average Wind power distribution of 2012")
ax[0, 2].set_title("Monthly average Wind power distribution of 2013")
ax[0, 3].set_title("Monthly average Wind power distribution of 2014")
ax[1, 0].set_title("Monthly average Wind power distribution of 2015")
ax[1, 1].set_title("Monthly average Wind power distribution of 2016")
ax[1, 2].set_title("Monthly average Wind power distribution of 2017")
ax[1, 3].set_title("Monthly average Wind power distribution of 2018")
ax[2, 0].set_title("Monthly average Wind power distribution of 2019")
ax[2, 1].set_title("Monthly average Wind power distribution of 2020")
ax[2, 2].set_title("Monthly average Wind power distribution of 2021")

plt.show()

Observations:

• These distributions do not seem to follow a normal distribution.

# Setting the time series column as index:
df.set_index('dt', inplace=True)

3) Preprocessing

The Class below has a method that can transform a dataset with a Time series structure into a dataset that can be used in a supervised manner.

# Proprocessing class 
class Preprocessamento:

    def timeseries_to_supervised(self, df, n_features, n_target):
        n_linhas = 0
        colunas_features = ['var(t - {})'.format(str(i)) for i in range(n_features, -1, -1) if i != 0]
        colunas_target = ['var(t)' if i==0 else 'var(t + {})'.format(str(i)) for i in range(0, n_target)]
        colunas_total = colunas_features + colunas_target
        lista=[]
        
        while n_linhas <= len(df) - n_target - n_features:
            quantidade_de_features_iteracao = df.iloc[n_linhas:n_linhas + n_features].values
            quantidade_de_target_iteracao = df.iloc[n_linhas + n_features: n_linhas + n_features + n_target]
            
            linha = np.concatenate([quantidade_de_features_iteracao, quantidade_de_target_iteracao], axis=0)
            linha_reshape = linha.reshape(1, -1)
            lista.append(linha_reshape[0])
    
            n_linhas += 1
        df_iter = pd.DataFrame(lista, columns=colunas_total)

        return df_iter

# Function that can plot several metrics:
def metricas(X_test, y_test, models):
    for name, model in models.items():
        if name == 'LSTM':
            y_pred = []
            for i in range(len(X_test)):
                X_test_linha = X_test[i, 0:]
                X_test_reshaped = X_test_linha.reshape(1, 1, len(X_test_linha))
                predicoes = model.predict(X_test_reshaped, batch_size=1, verbose=0)
                retorno = [x for x in predicoes[0]][0]
                y_pred.append(retorno)
            y_pred = np.array(y_pred)
        else:
            y_pred = model.predict(X_test)
        
        mse = mean_squared_error(y_test, y_pred)
        rmse = np.sqrt(mse)
        coef_pearson = r_regression(y_pred.reshape(-1, 1), y_test)[0]
        print(f'Mean squared error: {mse}')
        print(f'Root Mean squared error: {rmse}')
        print(f'Coef de pearson: {coef_pearson}')
        print('###########################################\n')

        return mse, rmse, coef_pearson

# Splitting into train an test datasets:
percentagem_treino_inicial = 0.8

limite_treino_inicial = int(len(df)*percentagem_treino_inicial)
df_train_inicial = df.iloc[0:limite_treino_inicial]
df_test = df.iloc[limite_treino_inicial:]

# Splitting into training and validation datasets:
percentagem_treino = 0.8

limite_treino = int(len(df_train_inicial)*percentagem_treino)
df_train = df_train_inicial.iloc[0:limite_treino]
df_val = df_train_inicial.iloc[limite_treino:]

# Intance of Preprocessing class:
prep_obj = Preprocessamento()

We will use a window of 20 and a horizon of 10. In other words, we are going to use 20 features to predict ten timesteps ahead in time. Furthemore, it is important to say that we will make a model for each timestep ahead, so it is going to be 10 models for each algorithm.

# Transforming the time series into a supervised problem:
n_features = 20
n_target = 10

df_train_supervised = prep_obj.timeseries_to_supervised(df_train, n_features, n_target)
df_val_supervised = prep_obj.timeseries_to_supervised(df_val, n_features, n_target)
df_test_supervised = prep_obj.timeseries_to_supervised(df_test, n_features, n_target)

# Some descriptive statistics about the features:
df_train_supervised.describe()

	var(t - 20)	var(t - 19)	var(t - 18)	var(t - 17)	var(t - 16)	var(t - 15)	var(t - 14)	var(t - 13)	var(t - 12)	var(t - 11)	...	var(t)	var(t + 1)	var(t + 2)	var(t + 3)	var(t + 4)	var(t + 5)	var(t + 6)	var(t + 7)	var(t + 8)	var(t + 9)
count	246732.000000	246732.000000	246732.000000	246732.000000	246732.000000	246732.000000	246732.000000	246732.000000	246732.000000	246732.000000	...	246732.000000	246732.00000	246732.000000	246732.000000	246732.000000	246732.000000	246732.000000	246732.000000	246732.000000	246732.000000
mean	2629.614760	2629.621063	2629.621858	2629.621876	2629.622045	2629.622477	2629.622653	2629.622223	2629.621149	2629.619584	...	2629.571150	2629.56493	2629.558805	2629.552966	2629.546922	2629.541040	2629.534696	2629.528298	2629.521042	2629.513345
std	2577.736081	2577.739905	2577.740590	2577.740607	2577.740760	2577.741166	2577.741341	2577.740898	2577.739783	2577.738152	...	2577.683505	2577.67614	2577.668791	2577.661796	2577.654572	2577.647439	2577.639970	2577.632878	2577.624891	2577.616327
min	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	...	0.000000	0.00000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000
25%	710.000000	710.000000	710.000000	710.000000	710.000000	710.000000	710.000000	710.000000	710.000000	710.000000	...	710.000000	710.00000	710.000000	710.000000	710.000000	710.000000	710.000000	710.000000	710.000000	710.000000
50%	1756.000000	1756.000000	1756.000000	1756.000000	1756.000000	1756.000000	1756.000000	1756.000000	1756.000000	1756.000000	...	1756.000000	1756.00000	1756.000000	1756.000000	1756.000000	1756.000000	1756.000000	1756.000000	1756.000000	1756.000000
75%	3720.270000	3720.385000	3720.385000	3720.385000	3720.385000	3720.385000	3720.385000	3720.385000	3720.385000	3720.385000	...	3720.385000	3720.38500	3720.385000	3720.385000	3720.385000	3720.385000	3720.385000	3720.385000	3720.385000	3720.385000
max	14266.070000	14266.070000	14266.070000	14266.070000	14266.070000	14266.070000	14266.070000	14266.070000	14266.070000	14266.070000	...	14266.070000	14266.07000	14266.070000	14266.070000	14266.070000	14266.070000	14266.070000	14266.070000	14266.070000	14266.070000

8 rows × 30 columns

Observations:

• All of the features have approxmately the same mean and Standard Deviation.

# Splitting into train and test fetures and targets:
X_train, y_train = df_train_supervised.values[:, 0:n_features], df_train_supervised.values[:, n_features:]
X_val, y_val = df_val_supervised.values[:, 0:n_features], df_val_supervised.values[:, n_features:]
X_test, y_test = df_test_supervised.values[:, 0:n_features], df_test_supervised.values[:, n_features:]

# Features shape:
print('Feature shapes:')
print(f'Training: {X_train.shape}')
print(f'Validation: {X_val.shape}')
print(f'Test: {X_test.shape}')
print('#########################\n')

# Target shape:
print('Target shapes:')
print(f'Training: {y_train.shape}')
print(f'Validation: {y_val.shape}')
print(f'Test: {y_test.shape}')

Feature shapes:
Training: (246732, 20)
Validation: (61662, 20)
Test: (77085, 20)
#########################

Target shapes:
Training: (246732, 10)
Validation: (61662, 10)
Test: (77085, 10)

3.1) Data transformation

3.1.1) Standard Scale

Definition:

• The Standard Scaler is a technique that rescales the distribution of a variable so that the mean of the observed sample is 0 and the standard deviation is 1. It is particularly useful for algorithms that rely on distance measures, such as K-means and K-nearest neighbors (KNN). Additionally, it is a recommended choice for algorithms based on neural networks.

OBS: Standard Scaler can perform slightly worst than the other transformations because it assumes that the data is normally distributed. However you can still standardize your data.

Matematical Definition:

$X_{new_{i}} = \frac{X_{i} - \hat{\mu}_{i}}{\sigma_{i}}$

• $\mu:$ Mean of the sample.
• $\sigma:$ Standard Deviation of the sample.

# Fitting a Standard Scaler object:
std_scaler = StandardScaler()
std_scaler.fit(X_train)

# Transforming all the sets:
X_train_std = std_scaler.transform(X_train) 
X_val_std = std_scaler.transform(X_val)
X_test_std = std_scaler.transform(X_test)

4) Fitting models

# Create nested directories: 
def make_directory(path):
    try:
        os.makedirs(path)
    except FileExistsError:
        print("File already exists!")

# Directory name:
directory = dt.now().strftime("%Y-%m-%d__%H_%M_%S")

4.1) Multilayer Neural Network

# Function that structure a simple neural network architecture:
def mlp_simples(device):

    with tf.device(device):
        model = tf.keras.Sequential([
            tf.keras.layers.Dense(100, activation='relu', input_shape=X_train.shape[1:]),
            tf.keras.layers.Dense(50, activation='relu'),
            tf.keras.layers.Dense(25, activation='relu'),
            tf.keras.layers.Dense(10, activation='relu'),
            tf.keras.layers.Dense(1)
        ])
        model.compile(optimizer='adam', loss='mean_squared_error', metrics=['mse'])
        
    return model

# Making a directory for MLP models:
path_modelos_mlp_simples = "modelos_mlp_simples/{}".format(directory)
make_directory(path_modelos_mlp_simples)

# Training a MLP:
for target in range(n_target):
    
    # EarlyStopping callback:
    earlystopping = tf.keras.callbacks.EarlyStopping(patience=10, restore_best_weights=True)
    checkpoint_mlp = tf.keras.callbacks.ModelCheckpoint('{}/model{}_mlp.h5'.format(path_modelos_mlp_simples, target))
    model_mlp = mlp_simples('/device:GPU:0')
    history_2 = model_mlp.fit(X_train_std, y_train[:, target], epochs=60,
                    validation_data=(X_val_std, y_val[:, target]), callbacks=[earlystopping, checkpoint_mlp])

File already exists!
Epoch 1/60
7711/7711 [==============================] - 14s 2ms/step - loss: 428830.5938 - mse: 428830.5938 - val_loss: 272890.6562 - val_mse: 272890.6562
Epoch 2/60
  60/7711 [..............................] - ETA: 13s - loss: 113575.3281 - mse: 113575.3281

c:\Users\caios\AppData\Local\Programs\Python\Python311\Lib\site-packages\keras\src\engine\training.py:3000: UserWarning:

You are saving your model as an HDF5 file via `model.save()`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')`.

7711/7711 [==============================] - 11s 1ms/step - loss: 146245.0156 - mse: 146245.0156 - val_loss: 286984.1250 - val_mse: 286984.1250
Epoch 3/60
7711/7711 [==============================] - 10s 1ms/step - loss: 145753.0781 - mse: 145753.0781 - val_loss: 266464.6250 - val_mse: 266464.6250
Epoch 4/60
7711/7711 [==============================] - 11s 1ms/step - loss: 145245.1875 - mse: 145245.1875 - val_loss: 267078.1875 - val_mse: 267078.1875
Epoch 5/60
7711/7711 [==============================] - 12s 2ms/step - loss: 144143.6406 - mse: 144143.6406 - val_loss: 261799.9062 - val_mse: 261799.9062
Epoch 6/60
7711/7711 [==============================] - 11s 1ms/step - loss: 142808.4688 - mse: 142808.4688 - val_loss: 262283.1562 - val_mse: 262283.1562
Epoch 7/60
7711/7711 [==============================] - 11s 1ms/step - loss: 141882.1719 - mse: 141882.1719 - val_loss: 256878.8438 - val_mse: 256878.8438
Epoch 8/60
7711/7711 [==============================] - 11s 1ms/step - loss: 139972.2656 - mse: 139972.2656 - val_loss: 258874.5469 - val_mse: 258874.5469
Epoch 9/60
7711/7711 [==============================] - 11s 1ms/step - loss: 139518.5312 - mse: 139518.5312 - val_loss: 253696.1094 - val_mse: 253696.1094
Epoch 10/60
7711/7711 [==============================] - 11s 1ms/step - loss: 138247.3438 - mse: 138247.3438 - val_loss: 252601.0781 - val_mse: 252601.0781
Epoch 11/60
7711/7711 [==============================] - 11s 1ms/step - loss: 137381.6719 - mse: 137381.6719 - val_loss: 252753.1094 - val_mse: 252753.1094
Epoch 12/60
7711/7711 [==============================] - 11s 1ms/step - loss: 136789.8281 - mse: 136789.8281 - val_loss: 335963.0312 - val_mse: 335963.0312
Epoch 13/60
7711/7711 [==============================] - 11s 1ms/step - loss: 136090.0156 - mse: 136090.0156 - val_loss: 246943.6562 - val_mse: 246943.6562
Epoch 14/60
7711/7711 [==============================] - 11s 1ms/step - loss: 134976.2188 - mse: 134976.2188 - val_loss: 246803.7969 - val_mse: 246803.7969
Epoch 15/60
7711/7711 [==============================] - 11s 1ms/step - loss: 134758.7656 - mse: 134758.7656 - val_loss: 260979.8906 - val_mse: 260979.8906
Epoch 16/60
7711/7711 [==============================] - 11s 1ms/step - loss: 134038.1094 - mse: 134038.1094 - val_loss: 244349.4062 - val_mse: 244349.4062
Epoch 17/60
7711/7711 [==============================] - 11s 1ms/step - loss: 133334.5625 - mse: 133334.5625 - val_loss: 247565.7969 - val_mse: 247565.7969
Epoch 18/60
7711/7711 [==============================] - 10s 1ms/step - loss: 132970.0781 - mse: 132970.0781 - val_loss: 251883.8438 - val_mse: 251883.8438
Epoch 19/60
7711/7711 [==============================] - 10s 1ms/step - loss: 132831.4688 - mse: 132831.4688 - val_loss: 246898.6094 - val_mse: 246898.6094
Epoch 20/60
7711/7711 [==============================] - 10s 1ms/step - loss: 132526.2969 - mse: 132526.2969 - val_loss: 241652.5312 - val_mse: 241652.5312
Epoch 21/60
7711/7711 [==============================] - 10s 1ms/step - loss: 132300.5781 - mse: 132300.5781 - val_loss: 245074.3594 - val_mse: 245074.3594
Epoch 22/60
7711/7711 [==============================] - 10s 1ms/step - loss: 132478.0938 - mse: 132478.0938 - val_loss: 241250.8594 - val_mse: 241250.8594
Epoch 23/60
7711/7711 [==============================] - 10s 1ms/step - loss: 132208.5312 - mse: 132208.5312 - val_loss: 253652.0000 - val_mse: 253652.0000
Epoch 24/60
7711/7711 [==============================] - 10s 1ms/step - loss: 131900.4531 - mse: 131900.4531 - val_loss: 246209.0156 - val_mse: 246209.0156
Epoch 25/60
7711/7711 [==============================] - 11s 1ms/step - loss: 131608.1094 - mse: 131608.1094 - val_loss: 240958.4531 - val_mse: 240958.4531
Epoch 26/60
7711/7711 [==============================] - 12s 2ms/step - loss: 131384.4219 - mse: 131384.4219 - val_loss: 244431.8594 - val_mse: 244431.8594
Epoch 27/60
7711/7711 [==============================] - 14s 2ms/step - loss: 131415.0469 - mse: 131415.0469 - val_loss: 239510.2969 - val_mse: 239510.2969
Epoch 28/60
7711/7711 [==============================] - 9s 1ms/step - loss: 131539.8594 - mse: 131539.8594 - val_loss: 245103.1406 - val_mse: 245103.1406
Epoch 29/60
7711/7711 [==============================] - 10s 1ms/step - loss: 130709.4375 - mse: 130709.4375 - val_loss: 241290.2969 - val_mse: 241290.2969
Epoch 30/60
7711/7711 [==============================] - 10s 1ms/step - loss: 131137.7500 - mse: 131137.7500 - val_loss: 241485.0938 - val_mse: 241485.0938
Epoch 31/60
7711/7711 [==============================] - 9s 1ms/step - loss: 130929.8438 - mse: 130929.8438 - val_loss: 238599.6250 - val_mse: 238599.6250
Epoch 32/60
7711/7711 [==============================] - 10s 1ms/step - loss: 130753.1562 - mse: 130753.1562 - val_loss: 252188.1875 - val_mse: 252188.1875
Epoch 33/60
7711/7711 [==============================] - 9s 1ms/step - loss: 130779.6562 - mse: 130779.6562 - val_loss: 241061.7969 - val_mse: 241061.7969
Epoch 34/60
7711/7711 [==============================] - 9s 1ms/step - loss: 130324.4375 - mse: 130324.4375 - val_loss: 239819.1875 - val_mse: 239819.1875
Epoch 35/60
7711/7711 [==============================] - 9s 1ms/step - loss: 130144.3047 - mse: 130144.3047 - val_loss: 274614.7188 - val_mse: 274614.7188
Epoch 36/60
7711/7711 [==============================] - 10s 1ms/step - loss: 130336.1250 - mse: 130336.1250 - val_loss: 245095.1250 - val_mse: 245095.1250
Epoch 37/60
7711/7711 [==============================] - 9s 1ms/step - loss: 130273.3516 - mse: 130273.3516 - val_loss: 243156.8594 - val_mse: 243156.8594
Epoch 38/60
7711/7711 [==============================] - 9s 1ms/step - loss: 130003.5859 - mse: 130003.5859 - val_loss: 252268.5625 - val_mse: 252268.5625
Epoch 39/60
7711/7711 [==============================] - 9s 1ms/step - loss: 130001.9375 - mse: 130001.9375 - val_loss: 244496.4062 - val_mse: 244496.4062
Epoch 40/60
7711/7711 [==============================] - 9s 1ms/step - loss: 129528.0078 - mse: 129528.0078 - val_loss: 239984.5000 - val_mse: 239984.5000
Epoch 41/60
7711/7711 [==============================] - 9s 1ms/step - loss: 129827.2656 - mse: 129827.2656 - val_loss: 238596.0312 - val_mse: 238596.0312
Epoch 42/60
7711/7711 [==============================] - 10s 1ms/step - loss: 130136.7266 - mse: 130136.7266 - val_loss: 236898.1094 - val_mse: 236898.1094
Epoch 43/60
7711/7711 [==============================] - 9s 1ms/step - loss: 129577.0391 - mse: 129577.0391 - val_loss: 312004.2500 - val_mse: 312004.2500
Epoch 44/60
7711/7711 [==============================] - 9s 1ms/step - loss: 129546.1328 - mse: 129546.1328 - val_loss: 263956.8125 - val_mse: 263956.8125
Epoch 45/60
7711/7711 [==============================] - 9s 1ms/step - loss: 129704.7188 - mse: 129704.7188 - val_loss: 237947.0625 - val_mse: 237947.0625
Epoch 46/60
7711/7711 [==============================] - 9s 1ms/step - loss: 129506.3672 - mse: 129506.3672 - val_loss: 249181.4688 - val_mse: 249181.4688
Epoch 47/60
7711/7711 [==============================] - 9s 1ms/step - loss: 128746.6719 - mse: 128746.6719 - val_loss: 235989.9844 - val_mse: 235989.9844
Epoch 48/60
7711/7711 [==============================] - 9s 1ms/step - loss: 129276.4531 - mse: 129276.4531 - val_loss: 257481.9062 - val_mse: 257481.9062
Epoch 49/60
7711/7711 [==============================] - 9s 1ms/step - loss: 129318.8906 - mse: 129318.8906 - val_loss: 235955.8438 - val_mse: 235955.8438
Epoch 50/60
7711/7711 [==============================] - 9s 1ms/step - loss: 128780.6094 - mse: 128780.6094 - val_loss: 238424.5312 - val_mse: 238424.5312
Epoch 51/60
7711/7711 [==============================] - 9s 1ms/step - loss: 128777.7969 - mse: 128777.7969 - val_loss: 237672.7500 - val_mse: 237672.7500
Epoch 52/60
7711/7711 [==============================] - 10s 1ms/step - loss: 128884.5469 - mse: 128884.5469 - val_loss: 241035.5781 - val_mse: 241035.5781
Epoch 53/60
7711/7711 [==============================] - 9s 1ms/step - loss: 128660.7031 - mse: 128660.7031 - val_loss: 242173.1719 - val_mse: 242173.1719
Epoch 54/60
7711/7711 [==============================] - 8s 1ms/step - loss: 128956.7422 - mse: 128956.7422 - val_loss: 238546.7656 - val_mse: 238546.7656
Epoch 55/60
7711/7711 [==============================] - 8s 1ms/step - loss: 128472.0000 - mse: 128472.0000 - val_loss: 238354.4531 - val_mse: 238354.4531
Epoch 56/60
7711/7711 [==============================] - 9s 1ms/step - loss: 128452.5547 - mse: 128452.5547 - val_loss: 240689.1250 - val_mse: 240689.1250
Epoch 57/60
7711/7711 [==============================] - 9s 1ms/step - loss: 128655.0703 - mse: 128655.0703 - val_loss: 239651.1250 - val_mse: 239651.1250
Epoch 58/60
7711/7711 [==============================] - 9s 1ms/step - loss: 128188.9297 - mse: 128188.9297 - val_loss: 239061.8125 - val_mse: 239061.8125
Epoch 59/60
7711/7711 [==============================] - 10s 1ms/step - loss: 128470.8359 - mse: 128470.8359 - val_loss: 236822.5938 - val_mse: 236822.5938
Epoch 1/60
7711/7711 [==============================] - 10s 1ms/step - loss: 494717.0312 - mse: 494717.0312 - val_loss: 406148.9375 - val_mse: 406148.9375
Epoch 2/60
7711/7711 [==============================] - 8s 1ms/step - loss: 218347.7344 - mse: 218347.7344 - val_loss: 417986.1562 - val_mse: 417986.1562
Epoch 3/60
7711/7711 [==============================] - 9s 1ms/step - loss: 215702.9688 - mse: 215702.9688 - val_loss: 394312.4375 - val_mse: 394312.4375
Epoch 4/60
7711/7711 [==============================] - 9s 1ms/step - loss: 212442.8906 - mse: 212442.8906 - val_loss: 405709.7812 - val_mse: 405709.7812
Epoch 5/60
7711/7711 [==============================] - 9s 1ms/step - loss: 209090.5781 - mse: 209090.5781 - val_loss: 376947.3750 - val_mse: 376947.3750
Epoch 6/60
7711/7711 [==============================] - 9s 1ms/step - loss: 206035.5469 - mse: 206035.5469 - val_loss: 368228.7812 - val_mse: 368228.7812
Epoch 7/60
7711/7711 [==============================] - 9s 1ms/step - loss: 203611.2344 - mse: 203611.2344 - val_loss: 450477.6875 - val_mse: 450477.6875
Epoch 8/60
7711/7711 [==============================] - 9s 1ms/step - loss: 202104.6875 - mse: 202104.6875 - val_loss: 384779.1250 - val_mse: 384779.1250
Epoch 9/60
7711/7711 [==============================] - 9s 1ms/step - loss: 201274.1875 - mse: 201274.1875 - val_loss: 361168.4375 - val_mse: 361168.4375
Epoch 10/60
7711/7711 [==============================] - 9s 1ms/step - loss: 200654.1250 - mse: 200654.1250 - val_loss: 364385.1875 - val_mse: 364385.1875
Epoch 11/60
7711/7711 [==============================] - 9s 1ms/step - loss: 199860.1719 - mse: 199860.1719 - val_loss: 360451.8438 - val_mse: 360451.8438
Epoch 12/60
7711/7711 [==============================] - 9s 1ms/step - loss: 199265.2500 - mse: 199265.2500 - val_loss: 361379.8125 - val_mse: 361379.8125
Epoch 13/60
7711/7711 [==============================] - 9s 1ms/step - loss: 198869.9375 - mse: 198869.9375 - val_loss: 356438.4062 - val_mse: 356438.4062
Epoch 14/60
7711/7711 [==============================] - 12s 2ms/step - loss: 198659.6875 - mse: 198659.6875 - val_loss: 393500.9688 - val_mse: 393500.9688
Epoch 15/60
7711/7711 [==============================] - 10s 1ms/step - loss: 198076.7500 - mse: 198076.7500 - val_loss: 360047.3438 - val_mse: 360047.3438
Epoch 16/60
7711/7711 [==============================] - 11s 1ms/step - loss: 197964.7812 - mse: 197964.7812 - val_loss: 440002.4375 - val_mse: 440002.4375
Epoch 17/60
7711/7711 [==============================] - 11s 1ms/step - loss: 197871.2031 - mse: 197871.2031 - val_loss: 353919.9688 - val_mse: 353919.9688
Epoch 18/60
7711/7711 [==============================] - 10s 1ms/step - loss: 197649.3906 - mse: 197649.3906 - val_loss: 356328.0000 - val_mse: 356328.0000
Epoch 19/60
7711/7711 [==============================] - 13s 2ms/step - loss: 197508.0625 - mse: 197508.0625 - val_loss: 363686.2812 - val_mse: 363686.2812
Epoch 20/60
7711/7711 [==============================] - 13s 2ms/step - loss: 197360.0156 - mse: 197360.0156 - val_loss: 363940.1562 - val_mse: 363940.1562
Epoch 21/60
7711/7711 [==============================] - 12s 2ms/step - loss: 197191.6094 - mse: 197191.6094 - val_loss: 356275.0000 - val_mse: 356275.0000
Epoch 22/60
7711/7711 [==============================] - 14s 2ms/step - loss: 196844.4844 - mse: 196844.4844 - val_loss: 366530.5938 - val_mse: 366530.5938
Epoch 23/60
7711/7711 [==============================] - 12s 2ms/step - loss: 196574.8594 - mse: 196574.8594 - val_loss: 354957.5938 - val_mse: 354957.5938
Epoch 24/60
7711/7711 [==============================] - 15s 2ms/step - loss: 196222.7812 - mse: 196222.7812 - val_loss: 352584.3125 - val_mse: 352584.3125
Epoch 25/60
7711/7711 [==============================] - 13s 2ms/step - loss: 196488.5938 - mse: 196488.5938 - val_loss: 358782.5000 - val_mse: 358782.5000
Epoch 26/60
7711/7711 [==============================] - 14s 2ms/step - loss: 196070.7031 - mse: 196070.7031 - val_loss: 352803.1562 - val_mse: 352803.1562
Epoch 27/60
7711/7711 [==============================] - 12s 2ms/step - loss: 196436.9219 - mse: 196436.9219 - val_loss: 367376.3438 - val_mse: 367376.3438
Epoch 28/60
7711/7711 [==============================] - 14s 2ms/step - loss: 196426.1719 - mse: 196426.1719 - val_loss: 359766.5938 - val_mse: 359766.5938
Epoch 29/60
7711/7711 [==============================] - 11s 1ms/step - loss: 195389.7656 - mse: 195389.7656 - val_loss: 354344.6562 - val_mse: 354344.6562
Epoch 30/60
7711/7711 [==============================] - 11s 1ms/step - loss: 195744.0625 - mse: 195744.0625 - val_loss: 352530.5938 - val_mse: 352530.5938
Epoch 31/60
7711/7711 [==============================] - 10s 1ms/step - loss: 195696.1094 - mse: 195696.1094 - val_loss: 364217.5000 - val_mse: 364217.5000
Epoch 32/60
7711/7711 [==============================] - 17s 2ms/step - loss: 194833.7500 - mse: 194833.7500 - val_loss: 386991.1250 - val_mse: 386991.1250
Epoch 33/60
7711/7711 [==============================] - 13s 2ms/step - loss: 195383.9062 - mse: 195383.9062 - val_loss: 418966.9062 - val_mse: 418966.9062
Epoch 34/60
7711/7711 [==============================] - 9s 1ms/step - loss: 194917.5000 - mse: 194917.5000 - val_loss: 351359.7812 - val_mse: 351359.7812
Epoch 35/60
7711/7711 [==============================] - 9s 1ms/step - loss: 195028.5781 - mse: 195028.5781 - val_loss: 350623.4375 - val_mse: 350623.4375
Epoch 36/60
7711/7711 [==============================] - 9s 1ms/step - loss: 194894.6406 - mse: 194894.6406 - val_loss: 388541.0625 - val_mse: 388541.0625
Epoch 37/60
7711/7711 [==============================] - 9s 1ms/step - loss: 194401.3438 - mse: 194401.3438 - val_loss: 351177.4062 - val_mse: 351177.4062
Epoch 38/60
7711/7711 [==============================] - 9s 1ms/step - loss: 194038.7344 - mse: 194038.7344 - val_loss: 356542.5000 - val_mse: 356542.5000
Epoch 39/60
7711/7711 [==============================] - 9s 1ms/step - loss: 194546.6719 - mse: 194546.6719 - val_loss: 360172.1250 - val_mse: 360172.1250
Epoch 40/60
7711/7711 [==============================] - 9s 1ms/step - loss: 194257.6094 - mse: 194257.6094 - val_loss: 353909.8125 - val_mse: 353909.8125
Epoch 41/60
7711/7711 [==============================] - 15s 2ms/step - loss: 194229.2031 - mse: 194229.2031 - val_loss: 351575.6250 - val_mse: 351575.6250
Epoch 42/60
7711/7711 [==============================] - 14s 2ms/step - loss: 194053.6719 - mse: 194053.6719 - val_loss: 349028.5000 - val_mse: 349028.5000
Epoch 43/60
7711/7711 [==============================] - 13s 2ms/step - loss: 193790.9062 - mse: 193790.9062 - val_loss: 355547.4688 - val_mse: 355547.4688
Epoch 44/60
7711/7711 [==============================] - 15s 2ms/step - loss: 193593.2188 - mse: 193593.2188 - val_loss: 360764.8438 - val_mse: 360764.8438
Epoch 45/60
7711/7711 [==============================] - 10s 1ms/step - loss: 193448.8750 - mse: 193448.8750 - val_loss: 350606.3125 - val_mse: 350606.3125
Epoch 46/60
7711/7711 [==============================] - 11s 1ms/step - loss: 193662.2969 - mse: 193662.2969 - val_loss: 350303.6562 - val_mse: 350303.6562
Epoch 47/60
7711/7711 [==============================] - 10s 1ms/step - loss: 193584.8750 - mse: 193584.8750 - val_loss: 351263.6875 - val_mse: 351263.6875
Epoch 48/60
7711/7711 [==============================] - 10s 1ms/step - loss: 193476.2500 - mse: 193476.2500 - val_loss: 348935.8438 - val_mse: 348935.8438
Epoch 49/60
7711/7711 [==============================] - 10s 1ms/step - loss: 193472.2812 - mse: 193472.2812 - val_loss: 350466.5938 - val_mse: 350466.5938
Epoch 50/60
7711/7711 [==============================] - 10s 1ms/step - loss: 193169.8906 - mse: 193169.8906 - val_loss: 360996.9375 - val_mse: 360996.9375
Epoch 51/60
7711/7711 [==============================] - 11s 1ms/step - loss: 192655.0469 - mse: 192655.0469 - val_loss: 350172.5312 - val_mse: 350172.5312
Epoch 52/60
7711/7711 [==============================] - 10s 1ms/step - loss: 193111.9531 - mse: 193111.9531 - val_loss: 355583.3750 - val_mse: 355583.3750
Epoch 53/60
7711/7711 [==============================] - 10s 1ms/step - loss: 192914.5625 - mse: 192914.5625 - val_loss: 347834.3438 - val_mse: 347834.3438
Epoch 54/60
7711/7711 [==============================] - 11s 1ms/step - loss: 192857.1875 - mse: 192857.1875 - val_loss: 370526.8125 - val_mse: 370526.8125
Epoch 55/60
7711/7711 [==============================] - 9s 1ms/step - loss: 192319.5625 - mse: 192319.5625 - val_loss: 350713.3438 - val_mse: 350713.3438
Epoch 56/60
7711/7711 [==============================] - 10s 1ms/step - loss: 192948.4844 - mse: 192948.4844 - val_loss: 353388.5625 - val_mse: 353388.5625
Epoch 57/60
7711/7711 [==============================] - 9s 1ms/step - loss: 192234.5938 - mse: 192234.5938 - val_loss: 347106.7812 - val_mse: 347106.7812
Epoch 58/60
7711/7711 [==============================] - 9s 1ms/step - loss: 192593.0312 - mse: 192593.0312 - val_loss: 348541.7500 - val_mse: 348541.7500
Epoch 59/60
7711/7711 [==============================] - 9s 1ms/step - loss: 192432.3281 - mse: 192432.3281 - val_loss: 358283.8750 - val_mse: 358283.8750
Epoch 60/60
7711/7711 [==============================] - 9s 1ms/step - loss: 192211.4375 - mse: 192211.4375 - val_loss: 363085.5000 - val_mse: 363085.5000
Epoch 1/60
7711/7711 [==============================] - 9s 1ms/step - loss: 543548.7500 - mse: 543548.7500 - val_loss: 548966.3125 - val_mse: 548966.3125
Epoch 2/60
7711/7711 [==============================] - 9s 1ms/step - loss: 298277.1562 - mse: 298277.1562 - val_loss: 630206.8750 - val_mse: 630206.8750
Epoch 3/60
7711/7711 [==============================] - 13s 2ms/step - loss: 294197.0312 - mse: 294197.0312 - val_loss: 530175.4375 - val_mse: 530175.4375
Epoch 4/60
7711/7711 [==============================] - 11s 1ms/step - loss: 288927.6875 - mse: 288927.6875 - val_loss: 531053.8750 - val_mse: 531053.8750
Epoch 5/60
7711/7711 [==============================] - 10s 1ms/step - loss: 283985.7188 - mse: 283985.7188 - val_loss: 534163.0000 - val_mse: 534163.0000
Epoch 6/60
7711/7711 [==============================] - 10s 1ms/step - loss: 280360.3750 - mse: 280360.3750 - val_loss: 556362.6875 - val_mse: 556362.6875
Epoch 7/60
7711/7711 [==============================] - 11s 1ms/step - loss: 278184.2188 - mse: 278184.2188 - val_loss: 524089.3125 - val_mse: 524089.3125
Epoch 8/60
7711/7711 [==============================] - 12s 2ms/step - loss: 275713.2188 - mse: 275713.2188 - val_loss: 525468.1875 - val_mse: 525468.1875
Epoch 9/60
7711/7711 [==============================] - 12s 2ms/step - loss: 274222.5938 - mse: 274222.5938 - val_loss: 497920.1562 - val_mse: 497920.1562
Epoch 10/60
7711/7711 [==============================] - 11s 1ms/step - loss: 273291.4688 - mse: 273291.4688 - val_loss: 495599.0938 - val_mse: 495599.0938
Epoch 11/60
7711/7711 [==============================] - 10s 1ms/step - loss: 272053.6875 - mse: 272053.6875 - val_loss: 570515.1875 - val_mse: 570515.1875
Epoch 12/60
7711/7711 [==============================] - 11s 1ms/step - loss: 271309.4375 - mse: 271309.4375 - val_loss: 528299.8125 - val_mse: 528299.8125
Epoch 13/60
7711/7711 [==============================] - 12s 2ms/step - loss: 270829.3438 - mse: 270829.3438 - val_loss: 498641.6875 - val_mse: 498641.6875
Epoch 14/60
7711/7711 [==============================] - 11s 1ms/step - loss: 270926.6875 - mse: 270926.6875 - val_loss: 491833.2500 - val_mse: 491833.2500
Epoch 15/60
7711/7711 [==============================] - 10s 1ms/step - loss: 270359.5000 - mse: 270359.5000 - val_loss: 490311.5312 - val_mse: 490311.5312
Epoch 16/60
7711/7711 [==============================] - 11s 1ms/step - loss: 269858.8125 - mse: 269858.8125 - val_loss: 487854.9688 - val_mse: 487854.9688
Epoch 17/60
7711/7711 [==============================] - 10s 1ms/step - loss: 269031.7812 - mse: 269031.7812 - val_loss: 485598.0938 - val_mse: 485598.0938
Epoch 18/60
7711/7711 [==============================] - 10s 1ms/step - loss: 269515.5938 - mse: 269515.5938 - val_loss: 613818.1875 - val_mse: 613818.1875
Epoch 19/60
7711/7711 [==============================] - 10s 1ms/step - loss: 268302.5938 - mse: 268302.5938 - val_loss: 489949.3438 - val_mse: 489949.3438
Epoch 20/60
7711/7711 [==============================] - 10s 1ms/step - loss: 267673.4688 - mse: 267673.4688 - val_loss: 497665.5312 - val_mse: 497665.5312
Epoch 21/60
7711/7711 [==============================] - 10s 1ms/step - loss: 267779.5625 - mse: 267779.5625 - val_loss: 491565.7500 - val_mse: 491565.7500
Epoch 22/60
7711/7711 [==============================] - 10s 1ms/step - loss: 267245.2812 - mse: 267245.2812 - val_loss: 485135.0000 - val_mse: 485135.0000
Epoch 23/60
7711/7711 [==============================] - 10s 1ms/step - loss: 267572.9375 - mse: 267572.9375 - val_loss: 488007.0000 - val_mse: 488007.0000
Epoch 24/60
7711/7711 [==============================] - 10s 1ms/step - loss: 266816.9688 - mse: 266816.9688 - val_loss: 500509.6562 - val_mse: 500509.6562
Epoch 25/60
7711/7711 [==============================] - 10s 1ms/step - loss: 266106.9062 - mse: 266106.9062 - val_loss: 480024.8438 - val_mse: 480024.8438
Epoch 26/60
7711/7711 [==============================] - 11s 1ms/step - loss: 266564.4688 - mse: 266564.4688 - val_loss: 485087.5625 - val_mse: 485087.5625
Epoch 27/60
7711/7711 [==============================] - 11s 1ms/step - loss: 265810.0312 - mse: 265810.0312 - val_loss: 496068.5000 - val_mse: 496068.5000
Epoch 28/60
7711/7711 [==============================] - 12s 2ms/step - loss: 265656.4375 - mse: 265656.4375 - val_loss: 479445.7500 - val_mse: 479445.7500
Epoch 29/60
7711/7711 [==============================] - 12s 2ms/step - loss: 265551.2188 - mse: 265551.2188 - val_loss: 486081.0938 - val_mse: 486081.0938
Epoch 30/60
7711/7711 [==============================] - 11s 1ms/step - loss: 265217.2812 - mse: 265217.2812 - val_loss: 477740.4688 - val_mse: 477740.4688
Epoch 31/60
7711/7711 [==============================] - 10s 1ms/step - loss: 265183.3750 - mse: 265183.3750 - val_loss: 478187.4062 - val_mse: 478187.4062
Epoch 32/60
7711/7711 [==============================] - 9s 1ms/step - loss: 264599.0312 - mse: 264599.0312 - val_loss: 498323.3438 - val_mse: 498323.3438
Epoch 33/60
7711/7711 [==============================] - 9s 1ms/step - loss: 264284.3438 - mse: 264284.3438 - val_loss: 505083.0938 - val_mse: 505083.0938
Epoch 34/60
7711/7711 [==============================] - 9s 1ms/step - loss: 264169.5312 - mse: 264169.5312 - val_loss: 481236.8125 - val_mse: 481236.8125
Epoch 35/60
7711/7711 [==============================] - 9s 1ms/step - loss: 264522.0625 - mse: 264522.0625 - val_loss: 620522.2500 - val_mse: 620522.2500
Epoch 36/60
7711/7711 [==============================] - 9s 1ms/step - loss: 263984.7500 - mse: 263984.7500 - val_loss: 478122.9062 - val_mse: 478122.9062
Epoch 37/60
7711/7711 [==============================] - 9s 1ms/step - loss: 264040.1250 - mse: 264040.1250 - val_loss: 481663.8125 - val_mse: 481663.8125
Epoch 38/60
7711/7711 [==============================] - 9s 1ms/step - loss: 263720.1250 - mse: 263720.1250 - val_loss: 474879.1562 - val_mse: 474879.1562
Epoch 39/60
7711/7711 [==============================] - 9s 1ms/step - loss: 263046.0625 - mse: 263046.0625 - val_loss: 538736.5000 - val_mse: 538736.5000
Epoch 40/60
7711/7711 [==============================] - 9s 1ms/step - loss: 263639.4375 - mse: 263639.4375 - val_loss: 480189.1562 - val_mse: 480189.1562
Epoch 41/60
7711/7711 [==============================] - 9s 1ms/step - loss: 263211.4062 - mse: 263211.4062 - val_loss: 475036.5938 - val_mse: 475036.5938
Epoch 42/60
7711/7711 [==============================] - 9s 1ms/step - loss: 263613.0938 - mse: 263613.0938 - val_loss: 487427.6562 - val_mse: 487427.6562
Epoch 43/60
7711/7711 [==============================] - 9s 1ms/step - loss: 263341.5938 - mse: 263341.5938 - val_loss: 478562.3438 - val_mse: 478562.3438
Epoch 44/60
7711/7711 [==============================] - 10s 1ms/step - loss: 262372.3438 - mse: 262372.3438 - val_loss: 478301.6875 - val_mse: 478301.6875
Epoch 45/60
7711/7711 [==============================] - 10s 1ms/step - loss: 262639.3438 - mse: 262639.3438 - val_loss: 500038.8438 - val_mse: 500038.8438
Epoch 46/60
7711/7711 [==============================] - 10s 1ms/step - loss: 262649.8438 - mse: 262649.8438 - val_loss: 490457.2188 - val_mse: 490457.2188
Epoch 47/60
7711/7711 [==============================] - 9s 1ms/step - loss: 262818.7188 - mse: 262818.7188 - val_loss: 478123.3750 - val_mse: 478123.3750
Epoch 48/60
7711/7711 [==============================] - 9s 1ms/step - loss: 262218.6250 - mse: 262218.6250 - val_loss: 495274.4375 - val_mse: 495274.4375
Epoch 1/60
7711/7711 [==============================] - 10s 1ms/step - loss: 640479.9375 - mse: 640479.9375 - val_loss: 705158.0000 - val_mse: 705158.0000
Epoch 2/60
7711/7711 [==============================] - 9s 1ms/step - loss: 381157.2500 - mse: 381157.2500 - val_loss: 690602.3125 - val_mse: 690602.3125
Epoch 3/60
7711/7711 [==============================] - 10s 1ms/step - loss: 372191.0312 - mse: 372191.0312 - val_loss: 679887.3750 - val_mse: 679887.3750
Epoch 4/60
7711/7711 [==============================] - 9s 1ms/step - loss: 364769.4688 - mse: 364769.4688 - val_loss: 725863.6875 - val_mse: 725863.6875
Epoch 5/60
7711/7711 [==============================] - 9s 1ms/step - loss: 359798.7500 - mse: 359798.7500 - val_loss: 700268.2500 - val_mse: 700268.2500
Epoch 6/60
7711/7711 [==============================] - 9s 1ms/step - loss: 355999.0000 - mse: 355999.0000 - val_loss: 643940.9375 - val_mse: 643940.9375
Epoch 7/60
7711/7711 [==============================] - 9s 1ms/step - loss: 354443.7188 - mse: 354443.7188 - val_loss: 653783.5000 - val_mse: 653783.5000
Epoch 8/60
7711/7711 [==============================] - 9s 1ms/step - loss: 352400.6250 - mse: 352400.6250 - val_loss: 635480.8750 - val_mse: 635480.8750
Epoch 9/60
7711/7711 [==============================] - 9s 1ms/step - loss: 351317.3125 - mse: 351317.3125 - val_loss: 645506.1250 - val_mse: 645506.1250
Epoch 10/60
7711/7711 [==============================] - 9s 1ms/step - loss: 349700.9688 - mse: 349700.9688 - val_loss: 632168.1875 - val_mse: 632168.1875
Epoch 11/60
7711/7711 [==============================] - 10s 1ms/step - loss: 348883.7812 - mse: 348883.7812 - val_loss: 639395.9375 - val_mse: 639395.9375
Epoch 12/60
7711/7711 [==============================] - 10s 1ms/step - loss: 347861.4062 - mse: 347861.4062 - val_loss: 633562.6250 - val_mse: 633562.6250
Epoch 13/60
7711/7711 [==============================] - 9s 1ms/step - loss: 347733.4688 - mse: 347733.4688 - val_loss: 645869.8125 - val_mse: 645869.8125
Epoch 14/60
7711/7711 [==============================] - 9s 1ms/step - loss: 346681.3750 - mse: 346681.3750 - val_loss: 642356.4375 - val_mse: 642356.4375
Epoch 15/60
7711/7711 [==============================] - 9s 1ms/step - loss: 346125.1250 - mse: 346125.1250 - val_loss: 636211.5625 - val_mse: 636211.5625
Epoch 16/60
7711/7711 [==============================] - 9s 1ms/step - loss: 344871.5000 - mse: 344871.5000 - val_loss: 630088.0625 - val_mse: 630088.0625
Epoch 17/60
7711/7711 [==============================] - 9s 1ms/step - loss: 345399.8750 - mse: 345399.8750 - val_loss: 620533.1250 - val_mse: 620533.1250
Epoch 18/60
7711/7711 [==============================] - 9s 1ms/step - loss: 344478.9375 - mse: 344478.9375 - val_loss: 640179.9375 - val_mse: 640179.9375
Epoch 19/60
7711/7711 [==============================] - 9s 1ms/step - loss: 343686.6250 - mse: 343686.6250 - val_loss: 655093.8125 - val_mse: 655093.8125
Epoch 20/60
7711/7711 [==============================] - 9s 1ms/step - loss: 343676.1562 - mse: 343676.1562 - val_loss: 617327.5625 - val_mse: 617327.5625
Epoch 21/60
7711/7711 [==============================] - 9s 1ms/step - loss: 343076.8750 - mse: 343076.8750 - val_loss: 625080.7500 - val_mse: 625080.7500
Epoch 22/60
7711/7711 [==============================] - 9s 1ms/step - loss: 342592.6875 - mse: 342592.6875 - val_loss: 620585.2500 - val_mse: 620585.2500
Epoch 23/60
7711/7711 [==============================] - 9s 1ms/step - loss: 342209.4062 - mse: 342209.4062 - val_loss: 630751.6875 - val_mse: 630751.6875
Epoch 24/60
7711/7711 [==============================] - 9s 1ms/step - loss: 341668.7500 - mse: 341668.7500 - val_loss: 720819.2500 - val_mse: 720819.2500
Epoch 25/60
7711/7711 [==============================] - 9s 1ms/step - loss: 341324.3125 - mse: 341324.3125 - val_loss: 615263.1875 - val_mse: 615263.1875
Epoch 26/60
7711/7711 [==============================] - 9s 1ms/step - loss: 340797.2812 - mse: 340797.2812 - val_loss: 615646.5000 - val_mse: 615646.5000
Epoch 27/60
7711/7711 [==============================] - 9s 1ms/step - loss: 341166.2812 - mse: 341166.2812 - val_loss: 616373.3125 - val_mse: 616373.3125
Epoch 28/60
7711/7711 [==============================] - 9s 1ms/step - loss: 340043.4375 - mse: 340043.4375 - val_loss: 610617.3125 - val_mse: 610617.3125
Epoch 29/60
7711/7711 [==============================] - 9s 1ms/step - loss: 339250.2812 - mse: 339250.2812 - val_loss: 621553.8750 - val_mse: 621553.8750
Epoch 30/60
7711/7711 [==============================] - 9s 1ms/step - loss: 340015.6562 - mse: 340015.6562 - val_loss: 609950.3750 - val_mse: 609950.3750
Epoch 31/60
7711/7711 [==============================] - 9s 1ms/step - loss: 340648.0312 - mse: 340648.0312 - val_loss: 619313.3125 - val_mse: 619313.3125
Epoch 32/60
7711/7711 [==============================] - 9s 1ms/step - loss: 340182.3750 - mse: 340182.3750 - val_loss: 616297.0000 - val_mse: 616297.0000
Epoch 33/60
7711/7711 [==============================] - 9s 1ms/step - loss: 339385.8750 - mse: 339385.8750 - val_loss: 631218.3125 - val_mse: 631218.3125
Epoch 34/60
7711/7711 [==============================] - 9s 1ms/step - loss: 338949.0938 - mse: 338949.0938 - val_loss: 613903.9375 - val_mse: 613903.9375
Epoch 35/60
7711/7711 [==============================] - 9s 1ms/step - loss: 338744.3438 - mse: 338744.3438 - val_loss: 620905.5625 - val_mse: 620905.5625
Epoch 36/60
7711/7711 [==============================] - 9s 1ms/step - loss: 337768.6250 - mse: 337768.6250 - val_loss: 612846.8750 - val_mse: 612846.8750
Epoch 37/60
7711/7711 [==============================] - 9s 1ms/step - loss: 338654.7812 - mse: 338654.7812 - val_loss: 617500.2500 - val_mse: 617500.2500
Epoch 38/60
7711/7711 [==============================] - 9s 1ms/step - loss: 338291.9375 - mse: 338291.9375 - val_loss: 623221.5625 - val_mse: 623221.5625
Epoch 39/60
7711/7711 [==============================] - 9s 1ms/step - loss: 338424.8750 - mse: 338424.8750 - val_loss: 609357.8125 - val_mse: 609357.8125
Epoch 40/60
7711/7711 [==============================] - 9s 1ms/step - loss: 337847.0000 - mse: 337847.0000 - val_loss: 627372.1875 - val_mse: 627372.1875
Epoch 41/60
7711/7711 [==============================] - 9s 1ms/step - loss: 338134.0625 - mse: 338134.0625 - val_loss: 614831.1250 - val_mse: 614831.1250
Epoch 42/60
7711/7711 [==============================] - 9s 1ms/step - loss: 337787.8125 - mse: 337787.8125 - val_loss: 647392.8750 - val_mse: 647392.8750
Epoch 43/60
7711/7711 [==============================] - 9s 1ms/step - loss: 337537.6875 - mse: 337537.6875 - val_loss: 609873.6250 - val_mse: 609873.6250
Epoch 44/60
7711/7711 [==============================] - 9s 1ms/step - loss: 337649.7500 - mse: 337649.7500 - val_loss: 605828.5000 - val_mse: 605828.5000
Epoch 45/60
7711/7711 [==============================] - 9s 1ms/step - loss: 336791.5938 - mse: 336791.5938 - val_loss: 616592.0000 - val_mse: 616592.0000
Epoch 46/60
7711/7711 [==============================] - 9s 1ms/step - loss: 337096.9688 - mse: 337096.9688 - val_loss: 616981.8750 - val_mse: 616981.8750
Epoch 47/60
7711/7711 [==============================] - 9s 1ms/step - loss: 336946.6562 - mse: 336946.6562 - val_loss: 611088.7500 - val_mse: 611088.7500
Epoch 48/60
7711/7711 [==============================] - 9s 1ms/step - loss: 336532.0000 - mse: 336532.0000 - val_loss: 607783.6875 - val_mse: 607783.6875
Epoch 49/60
7711/7711 [==============================] - 9s 1ms/step - loss: 336331.0000 - mse: 336331.0000 - val_loss: 616880.6250 - val_mse: 616880.6250
Epoch 50/60
7711/7711 [==============================] - 9s 1ms/step - loss: 336244.9062 - mse: 336244.9062 - val_loss: 607811.6875 - val_mse: 607811.6875
Epoch 51/60
7711/7711 [==============================] - 9s 1ms/step - loss: 336211.2188 - mse: 336211.2188 - val_loss: 611260.8750 - val_mse: 611260.8750
Epoch 52/60
7711/7711 [==============================] - 9s 1ms/step - loss: 335650.5938 - mse: 335650.5938 - val_loss: 640860.6875 - val_mse: 640860.6875
Epoch 53/60
7711/7711 [==============================] - 9s 1ms/step - loss: 336425.9062 - mse: 336425.9062 - val_loss: 613632.4375 - val_mse: 613632.4375
Epoch 54/60
7711/7711 [==============================] - 9s 1ms/step - loss: 335736.4375 - mse: 335736.4375 - val_loss: 656387.6875 - val_mse: 656387.6875
Epoch 1/60
7711/7711 [==============================] - 10s 1ms/step - loss: 778951.8125 - mse: 778951.8125 - val_loss: 897251.5000 - val_mse: 897251.5000
Epoch 2/60
7711/7711 [==============================] - 9s 1ms/step - loss: 476948.5938 - mse: 476948.5938 - val_loss: 884858.8750 - val_mse: 884858.8750
Epoch 3/60
7711/7711 [==============================] - 9s 1ms/step - loss: 474033.5312 - mse: 474033.5312 - val_loss: 886426.3750 - val_mse: 886426.3750
Epoch 4/60
7711/7711 [==============================] - 9s 1ms/step - loss: 468014.6562 - mse: 468014.6562 - val_loss: 861623.5000 - val_mse: 861623.5000
Epoch 5/60
7711/7711 [==============================] - 11s 1ms/step - loss: 459218.3750 - mse: 459218.3750 - val_loss: 842642.1250 - val_mse: 842642.1250
Epoch 6/60
7711/7711 [==============================] - 12s 2ms/step - loss: 452817.6250 - mse: 452817.6250 - val_loss: 831285.5625 - val_mse: 831285.5625
Epoch 7/60
7711/7711 [==============================] - 10s 1ms/step - loss: 446428.5625 - mse: 446428.5625 - val_loss: 818229.6875 - val_mse: 818229.6875
Epoch 8/60
7711/7711 [==============================] - 10s 1ms/step - loss: 441991.4375 - mse: 441991.4375 - val_loss: 805474.9375 - val_mse: 805474.9375
Epoch 9/60
7711/7711 [==============================] - 12s 2ms/step - loss: 438967.1562 - mse: 438967.1562 - val_loss: 795317.5000 - val_mse: 795317.5000
Epoch 10/60
7711/7711 [==============================] - 12s 2ms/step - loss: 436377.5938 - mse: 436377.5938 - val_loss: 863349.1875 - val_mse: 863349.1875
Epoch 11/60
7711/7711 [==============================] - 11s 1ms/step - loss: 434758.3750 - mse: 434758.3750 - val_loss: 803879.6250 - val_mse: 803879.6250
Epoch 12/60
7711/7711 [==============================] - 11s 1ms/step - loss: 432512.9062 - mse: 432512.9062 - val_loss: 786905.3750 - val_mse: 786905.3750
Epoch 13/60
7711/7711 [==============================] - 10s 1ms/step - loss: 431797.0312 - mse: 431797.0312 - val_loss: 810194.8750 - val_mse: 810194.8750
Epoch 14/60
7711/7711 [==============================] - 11s 1ms/step - loss: 430791.3125 - mse: 430791.3125 - val_loss: 784365.9375 - val_mse: 784365.9375
Epoch 15/60
7711/7711 [==============================] - 11s 1ms/step - loss: 430534.0625 - mse: 430534.0625 - val_loss: 829775.5000 - val_mse: 829775.5000
Epoch 16/60
7711/7711 [==============================] - 10s 1ms/step - loss: 429933.4688 - mse: 429933.4688 - val_loss: 784172.1875 - val_mse: 784172.1875
Epoch 17/60
7711/7711 [==============================] - 10s 1ms/step - loss: 429233.7812 - mse: 429233.7812 - val_loss: 782226.3750 - val_mse: 782226.3750
Epoch 18/60
7711/7711 [==============================] - 10s 1ms/step - loss: 429233.5625 - mse: 429233.5625 - val_loss: 784080.2500 - val_mse: 784080.2500
Epoch 19/60
7711/7711 [==============================] - 11s 1ms/step - loss: 428533.0625 - mse: 428533.0625 - val_loss: 775613.0000 - val_mse: 775613.0000
Epoch 20/60
7711/7711 [==============================] - 11s 1ms/step - loss: 427382.1250 - mse: 427382.1250 - val_loss: 905891.5625 - val_mse: 905891.5625
Epoch 21/60
7711/7711 [==============================] - 11s 1ms/step - loss: 426537.4688 - mse: 426537.4688 - val_loss: 784542.6250 - val_mse: 784542.6250
Epoch 22/60
7711/7711 [==============================] - 10s 1ms/step - loss: 426059.0000 - mse: 426059.0000 - val_loss: 781943.0000 - val_mse: 781943.0000
Epoch 23/60
7711/7711 [==============================] - 13s 2ms/step - loss: 426361.3438 - mse: 426361.3438 - val_loss: 767673.0000 - val_mse: 767673.0000
Epoch 24/60
7711/7711 [==============================] - 11s 1ms/step - loss: 425075.5625 - mse: 425075.5625 - val_loss: 787082.9375 - val_mse: 787082.9375
Epoch 25/60
7711/7711 [==============================] - 12s 2ms/step - loss: 425295.6562 - mse: 425295.6562 - val_loss: 879157.6250 - val_mse: 879157.6250
Epoch 26/60
7711/7711 [==============================] - 12s 2ms/step - loss: 424687.4062 - mse: 424687.4062 - val_loss: 939933.9375 - val_mse: 939933.9375
Epoch 27/60
7711/7711 [==============================] - 11s 1ms/step - loss: 424104.9375 - mse: 424104.9375 - val_loss: 779014.0625 - val_mse: 779014.0625
Epoch 28/60
7711/7711 [==============================] - 11s 1ms/step - loss: 423708.1562 - mse: 423708.1562 - val_loss: 767648.8750 - val_mse: 767648.8750
Epoch 29/60
7711/7711 [==============================] - 12s 2ms/step - loss: 423013.2812 - mse: 423013.2812 - val_loss: 778856.1250 - val_mse: 778856.1250
Epoch 30/60
7711/7711 [==============================] - 12s 2ms/step - loss: 422649.5000 - mse: 422649.5000 - val_loss: 766976.6875 - val_mse: 766976.6875
Epoch 31/60
7711/7711 [==============================] - 11s 1ms/step - loss: 422233.9688 - mse: 422233.9688 - val_loss: 777825.3750 - val_mse: 777825.3750
Epoch 32/60
7711/7711 [==============================] - 10s 1ms/step - loss: 422135.3125 - mse: 422135.3125 - val_loss: 791503.2500 - val_mse: 791503.2500
Epoch 33/60
7711/7711 [==============================] - 12s 1ms/step - loss: 421662.0938 - mse: 421662.0938 - val_loss: 772355.0625 - val_mse: 772355.0625
Epoch 34/60
7711/7711 [==============================] - 11s 1ms/step - loss: 420954.8125 - mse: 420954.8125 - val_loss: 780982.7500 - val_mse: 780982.7500
Epoch 35/60
7711/7711 [==============================] - 11s 1ms/step - loss: 420382.0625 - mse: 420382.0625 - val_loss: 765014.8125 - val_mse: 765014.8125
Epoch 36/60
7711/7711 [==============================] - 10s 1ms/step - loss: 421087.4375 - mse: 421087.4375 - val_loss: 794569.3125 - val_mse: 794569.3125
Epoch 37/60
7711/7711 [==============================] - 10s 1ms/step - loss: 421592.8750 - mse: 421592.8750 - val_loss: 764055.4375 - val_mse: 764055.4375
Epoch 38/60
7711/7711 [==============================] - 10s 1ms/step - loss: 420526.1250 - mse: 420526.1250 - val_loss: 758052.3125 - val_mse: 758052.3125
Epoch 39/60
7711/7711 [==============================] - 11s 1ms/step - loss: 420413.6875 - mse: 420413.6875 - val_loss: 783448.1875 - val_mse: 783448.1875
Epoch 40/60
7711/7711 [==============================] - 11s 1ms/step - loss: 419646.3750 - mse: 419646.3750 - val_loss: 762208.1250 - val_mse: 762208.1250
Epoch 41/60
7711/7711 [==============================] - 10s 1ms/step - loss: 418734.1250 - mse: 418734.1250 - val_loss: 756886.4375 - val_mse: 756886.4375
Epoch 42/60
7711/7711 [==============================] - 10s 1ms/step - loss: 419354.2812 - mse: 419354.2812 - val_loss: 764782.8750 - val_mse: 764782.8750
Epoch 43/60
7711/7711 [==============================] - 11s 1ms/step - loss: 418709.6562 - mse: 418709.6562 - val_loss: 824495.2500 - val_mse: 824495.2500
Epoch 44/60
7711/7711 [==============================] - 14s 2ms/step - loss: 418657.6250 - mse: 418657.6250 - val_loss: 768467.7500 - val_mse: 768467.7500
Epoch 45/60
7711/7711 [==============================] - 11s 1ms/step - loss: 418869.1250 - mse: 418869.1250 - val_loss: 774622.3125 - val_mse: 774622.3125
Epoch 46/60
7711/7711 [==============================] - 11s 1ms/step - loss: 418472.8438 - mse: 418472.8438 - val_loss: 956475.8125 - val_mse: 956475.8125
Epoch 47/60
7711/7711 [==============================] - 11s 1ms/step - loss: 418372.7188 - mse: 418372.7188 - val_loss: 758789.3750 - val_mse: 758789.3750
Epoch 48/60
7711/7711 [==============================] - 11s 1ms/step - loss: 418974.2500 - mse: 418974.2500 - val_loss: 763732.0625 - val_mse: 763732.0625
Epoch 49/60
7711/7711 [==============================] - 12s 2ms/step - loss: 418257.7500 - mse: 418257.7500 - val_loss: 762854.0000 - val_mse: 762854.0000
Epoch 50/60
7711/7711 [==============================] - 11s 1ms/step - loss: 417721.7500 - mse: 417721.7500 - val_loss: 771228.9375 - val_mse: 771228.9375
Epoch 51/60
7711/7711 [==============================] - 10s 1ms/step - loss: 418078.6875 - mse: 418078.6875 - val_loss: 824662.3125 - val_mse: 824662.3125
Epoch 1/60
7711/7711 [==============================] - 13s 2ms/step - loss: 822089.8125 - mse: 822089.8125 - val_loss: 1092665.5000 - val_mse: 1092665.5000
Epoch 2/60
7711/7711 [==============================] - 15s 2ms/step - loss: 573867.0000 - mse: 573867.0000 - val_loss: 1073554.7500 - val_mse: 1073554.7500
Epoch 3/60
7711/7711 [==============================] - 10s 1ms/step - loss: 569243.3750 - mse: 569243.3750 - val_loss: 1076663.8750 - val_mse: 1076663.8750
Epoch 4/60
7711/7711 [==============================] - 10s 1ms/step - loss: 560329.1250 - mse: 560329.1250 - val_loss: 1057593.1250 - val_mse: 1057593.1250
Epoch 5/60
7711/7711 [==============================] - 10s 1ms/step - loss: 546283.9375 - mse: 546283.9375 - val_loss: 1270365.6250 - val_mse: 1270365.6250
Epoch 6/60
7711/7711 [==============================] - 10s 1ms/step - loss: 537370.0000 - mse: 537370.0000 - val_loss: 1067165.1250 - val_mse: 1067165.1250
Epoch 7/60
7711/7711 [==============================] - 11s 1ms/step - loss: 532168.6250 - mse: 532168.6250 - val_loss: 990216.7500 - val_mse: 990216.7500
Epoch 8/60
7711/7711 [==============================] - 10s 1ms/step - loss: 527877.5625 - mse: 527877.5625 - val_loss: 984519.7500 - val_mse: 984519.7500
Epoch 9/60
7711/7711 [==============================] - 10s 1ms/step - loss: 524842.0625 - mse: 524842.0625 - val_loss: 1027318.2500 - val_mse: 1027318.2500
Epoch 10/60
7711/7711 [==============================] - 10s 1ms/step - loss: 524189.3750 - mse: 524189.3750 - val_loss: 981241.9375 - val_mse: 981241.9375
Epoch 11/60
7711/7711 [==============================] - 10s 1ms/step - loss: 521482.3438 - mse: 521482.3438 - val_loss: 1026980.6250 - val_mse: 1026980.6250
Epoch 12/60
7711/7711 [==============================] - 10s 1ms/step - loss: 521400.5312 - mse: 521400.5312 - val_loss: 960179.6250 - val_mse: 960179.6250
Epoch 13/60
7711/7711 [==============================] - 10s 1ms/step - loss: 520923.3750 - mse: 520923.3750 - val_loss: 956437.1250 - val_mse: 956437.1250
Epoch 14/60
7711/7711 [==============================] - 11s 1ms/step - loss: 520810.4688 - mse: 520810.4688 - val_loss: 973217.1875 - val_mse: 973217.1875
Epoch 15/60
7711/7711 [==============================] - 11s 1ms/step - loss: 518788.3125 - mse: 518788.3125 - val_loss: 957400.4375 - val_mse: 957400.4375
Epoch 16/60
7711/7711 [==============================] - 11s 1ms/step - loss: 517376.2188 - mse: 517376.2188 - val_loss: 996320.4375 - val_mse: 996320.4375
Epoch 17/60
7711/7711 [==============================] - 10s 1ms/step - loss: 516700.4688 - mse: 516700.4688 - val_loss: 977813.6875 - val_mse: 977813.6875
Epoch 18/60
7711/7711 [==============================] - 10s 1ms/step - loss: 516091.7812 - mse: 516091.7812 - val_loss: 945316.6875 - val_mse: 945316.6875
Epoch 19/60
7711/7711 [==============================] - 10s 1ms/step - loss: 515172.4375 - mse: 515172.4375 - val_loss: 944262.5625 - val_mse: 944262.5625
Epoch 20/60
7711/7711 [==============================] - 10s 1ms/step - loss: 515676.5312 - mse: 515676.5312 - val_loss: 956029.8125 - val_mse: 956029.8125
Epoch 21/60
7711/7711 [==============================] - 10s 1ms/step - loss: 514419.7188 - mse: 514419.7188 - val_loss: 953953.3750 - val_mse: 953953.3750
Epoch 22/60
7711/7711 [==============================] - 10s 1ms/step - loss: 513857.2812 - mse: 513857.2812 - val_loss: 960057.1250 - val_mse: 960057.1250
Epoch 23/60
7711/7711 [==============================] - 10s 1ms/step - loss: 513526.7188 - mse: 513526.7188 - val_loss: 957107.3750 - val_mse: 957107.3750
Epoch 24/60
7711/7711 [==============================] - 10s 1ms/step - loss: 512539.0312 - mse: 512539.0312 - val_loss: 1130272.6250 - val_mse: 1130272.6250
Epoch 25/60
7711/7711 [==============================] - 10s 1ms/step - loss: 512385.6562 - mse: 512385.6562 - val_loss: 955762.7500 - val_mse: 955762.7500
Epoch 26/60
7711/7711 [==============================] - 10s 1ms/step - loss: 511086.6875 - mse: 511086.6875 - val_loss: 1033774.3125 - val_mse: 1033774.3125
Epoch 27/60
7711/7711 [==============================] - 10s 1ms/step - loss: 511227.5000 - mse: 511227.5000 - val_loss: 943798.3750 - val_mse: 943798.3750
Epoch 28/60
7711/7711 [==============================] - 11s 1ms/step - loss: 509917.4375 - mse: 509917.4375 - val_loss: 956527.3750 - val_mse: 956527.3750
Epoch 29/60
7711/7711 [==============================] - 11s 1ms/step - loss: 510419.4375 - mse: 510419.4375 - val_loss: 948150.4375 - val_mse: 948150.4375
Epoch 30/60
7711/7711 [==============================] - 10s 1ms/step - loss: 509777.9688 - mse: 509777.9688 - val_loss: 951321.6875 - val_mse: 951321.6875
Epoch 31/60
7711/7711 [==============================] - 10s 1ms/step - loss: 509850.0312 - mse: 509850.0312 - val_loss: 953499.1250 - val_mse: 953499.1250
Epoch 32/60
7711/7711 [==============================] - 11s 1ms/step - loss: 508404.5625 - mse: 508404.5625 - val_loss: 942554.5000 - val_mse: 942554.5000
Epoch 33/60
7711/7711 [==============================] - 10s 1ms/step - loss: 508642.7188 - mse: 508642.7188 - val_loss: 939956.3125 - val_mse: 939956.3125
Epoch 34/60
7711/7711 [==============================] - 11s 1ms/step - loss: 507190.8438 - mse: 507190.8438 - val_loss: 933004.8125 - val_mse: 933004.8125
Epoch 35/60
7711/7711 [==============================] - 11s 1ms/step - loss: 506557.5000 - mse: 506557.5000 - val_loss: 975317.6250 - val_mse: 975317.6250
Epoch 36/60
7711/7711 [==============================] - 11s 1ms/step - loss: 506210.5000 - mse: 506210.5000 - val_loss: 935929.6250 - val_mse: 935929.6250
Epoch 37/60
7711/7711 [==============================] - 10s 1ms/step - loss: 506076.0000 - mse: 506076.0000 - val_loss: 958417.9375 - val_mse: 958417.9375
Epoch 38/60
7711/7711 [==============================] - 14s 2ms/step - loss: 505795.8750 - mse: 505795.8750 - val_loss: 1027453.3125 - val_mse: 1027453.3125
Epoch 39/60
7711/7711 [==============================] - 17s 2ms/step - loss: 504930.6875 - mse: 504930.6875 - val_loss: 954653.4375 - val_mse: 954653.4375
Epoch 40/60
7711/7711 [==============================] - 10s 1ms/step - loss: 505140.3750 - mse: 505140.3750 - val_loss: 964011.3125 - val_mse: 964011.3125
Epoch 41/60
7711/7711 [==============================] - 9s 1ms/step - loss: 504096.3125 - mse: 504096.3125 - val_loss: 947568.0625 - val_mse: 947568.0625
Epoch 42/60
7711/7711 [==============================] - 11s 1ms/step - loss: 505114.3438 - mse: 505114.3438 - val_loss: 925723.4375 - val_mse: 925723.4375
Epoch 43/60
7711/7711 [==============================] - 9s 1ms/step - loss: 503383.2188 - mse: 503383.2188 - val_loss: 958306.4375 - val_mse: 958306.4375
Epoch 44/60
7711/7711 [==============================] - 9s 1ms/step - loss: 503377.9062 - mse: 503377.9062 - val_loss: 924451.5000 - val_mse: 924451.5000
Epoch 45/60
7711/7711 [==============================] - 9s 1ms/step - loss: 504144.8125 - mse: 504144.8125 - val_loss: 936101.1875 - val_mse: 936101.1875
Epoch 46/60
7711/7711 [==============================] - 10s 1ms/step - loss: 502474.4688 - mse: 502474.4688 - val_loss: 927120.9375 - val_mse: 927120.9375
Epoch 47/60
7711/7711 [==============================] - 9s 1ms/step - loss: 502370.3750 - mse: 502370.3750 - val_loss: 936023.3125 - val_mse: 936023.3125
Epoch 48/60
7711/7711 [==============================] - 9s 1ms/step - loss: 502839.3125 - mse: 502839.3125 - val_loss: 929154.8125 - val_mse: 929154.8125
Epoch 49/60
7711/7711 [==============================] - 15s 2ms/step - loss: 501862.8750 - mse: 501862.8750 - val_loss: 930205.5000 - val_mse: 930205.5000
Epoch 50/60
7711/7711 [==============================] - 11s 1ms/step - loss: 501642.9062 - mse: 501642.9062 - val_loss: 959624.9375 - val_mse: 959624.9375
Epoch 51/60
7711/7711 [==============================] - 10s 1ms/step - loss: 501601.1562 - mse: 501601.1562 - val_loss: 970214.5625 - val_mse: 970214.5625
Epoch 52/60
7711/7711 [==============================] - 9s 1ms/step - loss: 500898.8125 - mse: 500898.8125 - val_loss: 924499.7500 - val_mse: 924499.7500
Epoch 53/60
7711/7711 [==============================] - 9s 1ms/step - loss: 501057.6562 - mse: 501057.6562 - val_loss: 1087453.6250 - val_mse: 1087453.6250
Epoch 54/60
7711/7711 [==============================] - 9s 1ms/step - loss: 500612.3750 - mse: 500612.3750 - val_loss: 930046.1875 - val_mse: 930046.1875
Epoch 1/60
7711/7711 [==============================] - 10s 1ms/step - loss: 981734.0625 - mse: 981734.0625 - val_loss: 1339244.5000 - val_mse: 1339244.5000
Epoch 2/60
7711/7711 [==============================] - 9s 1ms/step - loss: 675785.6250 - mse: 675785.6250 - val_loss: 1294977.3750 - val_mse: 1294977.3750
Epoch 3/60
7711/7711 [==============================] - 9s 1ms/step - loss: 669547.6875 - mse: 669547.6875 - val_loss: 1274326.6250 - val_mse: 1274326.6250
Epoch 4/60
7711/7711 [==============================] - 9s 1ms/step - loss: 662412.8750 - mse: 662412.8750 - val_loss: 1313314.2500 - val_mse: 1313314.2500
Epoch 5/60
7711/7711 [==============================] - 9s 1ms/step - loss: 653098.8125 - mse: 653098.8125 - val_loss: 1232527.1250 - val_mse: 1232527.1250
Epoch 6/60
7711/7711 [==============================] - 9s 1ms/step - loss: 643368.3125 - mse: 643368.3125 - val_loss: 1264368.3750 - val_mse: 1264368.3750
Epoch 7/60
7711/7711 [==============================] - 9s 1ms/step - loss: 634029.1875 - mse: 634029.1875 - val_loss: 1208441.2500 - val_mse: 1208441.2500
Epoch 8/60
7711/7711 [==============================] - 9s 1ms/step - loss: 629700.1250 - mse: 629700.1250 - val_loss: 1173558.1250 - val_mse: 1173558.1250
Epoch 9/60
7711/7711 [==============================] - 9s 1ms/step - loss: 624610.1250 - mse: 624610.1250 - val_loss: 1178656.5000 - val_mse: 1178656.5000
Epoch 10/60
7711/7711 [==============================] - 9s 1ms/step - loss: 621151.3125 - mse: 621151.3125 - val_loss: 1163886.1250 - val_mse: 1163886.1250
Epoch 11/60
7711/7711 [==============================] - 9s 1ms/step - loss: 618601.5000 - mse: 618601.5000 - val_loss: 1331934.0000 - val_mse: 1331934.0000
Epoch 12/60
7711/7711 [==============================] - 13s 2ms/step - loss: 616586.5000 - mse: 616586.5000 - val_loss: 1175546.8750 - val_mse: 1175546.8750
Epoch 13/60
7711/7711 [==============================] - 10s 1ms/step - loss: 614787.3125 - mse: 614787.3125 - val_loss: 1174948.7500 - val_mse: 1174948.7500
Epoch 14/60
7711/7711 [==============================] - 9s 1ms/step - loss: 613054.5000 - mse: 613054.5000 - val_loss: 1173044.7500 - val_mse: 1173044.7500
Epoch 15/60
7711/7711 [==============================] - 9s 1ms/step - loss: 612301.8750 - mse: 612301.8750 - val_loss: 1143026.0000 - val_mse: 1143026.0000
Epoch 16/60
7711/7711 [==============================] - 10s 1ms/step - loss: 612083.3750 - mse: 612083.3750 - val_loss: 1148925.0000 - val_mse: 1148925.0000
Epoch 17/60
7711/7711 [==============================] - 9s 1ms/step - loss: 611215.9375 - mse: 611215.9375 - val_loss: 1147968.2500 - val_mse: 1147968.2500
Epoch 18/60
7711/7711 [==============================] - 9s 1ms/step - loss: 609028.1250 - mse: 609028.1250 - val_loss: 1126155.6250 - val_mse: 1126155.6250
Epoch 19/60
7711/7711 [==============================] - 9s 1ms/step - loss: 609474.3750 - mse: 609474.3750 - val_loss: 1166548.5000 - val_mse: 1166548.5000
Epoch 20/60
7711/7711 [==============================] - 9s 1ms/step - loss: 607584.1875 - mse: 607584.1875 - val_loss: 1137951.5000 - val_mse: 1137951.5000
Epoch 21/60
7711/7711 [==============================] - 9s 1ms/step - loss: 607689.5625 - mse: 607689.5625 - val_loss: 1156014.5000 - val_mse: 1156014.5000
Epoch 22/60
7711/7711 [==============================] - 9s 1ms/step - loss: 607546.6250 - mse: 607546.6250 - val_loss: 1142106.1250 - val_mse: 1142106.1250
Epoch 23/60
7711/7711 [==============================] - 9s 1ms/step - loss: 606734.0000 - mse: 606734.0000 - val_loss: 1128023.5000 - val_mse: 1128023.5000
Epoch 24/60
7711/7711 [==============================] - 9s 1ms/step - loss: 606110.5000 - mse: 606110.5000 - val_loss: 1144863.8750 - val_mse: 1144863.8750
Epoch 25/60
7711/7711 [==============================] - 9s 1ms/step - loss: 604953.2500 - mse: 604953.2500 - val_loss: 1160347.6250 - val_mse: 1160347.6250
Epoch 26/60
7711/7711 [==============================] - 9s 1ms/step - loss: 604546.4375 - mse: 604546.4375 - val_loss: 1121827.3750 - val_mse: 1121827.3750
Epoch 27/60
7711/7711 [==============================] - 9s 1ms/step - loss: 604484.0000 - mse: 604484.0000 - val_loss: 1120987.3750 - val_mse: 1120987.3750
Epoch 28/60
7711/7711 [==============================] - 9s 1ms/step - loss: 604431.8125 - mse: 604431.8125 - val_loss: 1124276.7500 - val_mse: 1124276.7500
Epoch 29/60
7711/7711 [==============================] - 9s 1ms/step - loss: 603637.5000 - mse: 603637.5000 - val_loss: 1128026.8750 - val_mse: 1128026.8750
Epoch 30/60
7711/7711 [==============================] - 11s 1ms/step - loss: 602685.1250 - mse: 602685.1250 - val_loss: 1115265.2500 - val_mse: 1115265.2500
Epoch 31/60
7711/7711 [==============================] - 9s 1ms/step - loss: 602495.5625 - mse: 602495.5625 - val_loss: 1128574.3750 - val_mse: 1128574.3750
Epoch 32/60
7711/7711 [==============================] - 9s 1ms/step - loss: 600988.1875 - mse: 600988.1875 - val_loss: 1152490.8750 - val_mse: 1152490.8750
Epoch 33/60
7711/7711 [==============================] - 10s 1ms/step - loss: 601937.0000 - mse: 601937.0000 - val_loss: 1142351.7500 - val_mse: 1142351.7500
Epoch 34/60
7711/7711 [==============================] - 12s 2ms/step - loss: 601274.7500 - mse: 601274.7500 - val_loss: 1151511.0000 - val_mse: 1151511.0000
Epoch 35/60
7711/7711 [==============================] - 13s 2ms/step - loss: 600328.8125 - mse: 600328.8125 - val_loss: 1110999.6250 - val_mse: 1110999.6250
Epoch 36/60
7711/7711 [==============================] - 9s 1ms/step - loss: 601184.0000 - mse: 601184.0000 - val_loss: 1162446.2500 - val_mse: 1162446.2500
Epoch 37/60
7711/7711 [==============================] - 9s 1ms/step - loss: 599595.7500 - mse: 599595.7500 - val_loss: 1146517.8750 - val_mse: 1146517.8750
Epoch 38/60
7711/7711 [==============================] - 9s 1ms/step - loss: 598719.0000 - mse: 598719.0000 - val_loss: 1106179.5000 - val_mse: 1106179.5000
Epoch 39/60
7711/7711 [==============================] - 9s 1ms/step - loss: 598833.5000 - mse: 598833.5000 - val_loss: 1120472.5000 - val_mse: 1120472.5000
Epoch 40/60
7711/7711 [==============================] - 9s 1ms/step - loss: 599331.8750 - mse: 599331.8750 - val_loss: 1122754.2500 - val_mse: 1122754.2500
Epoch 41/60
7711/7711 [==============================] - 9s 1ms/step - loss: 599445.8125 - mse: 599445.8125 - val_loss: 1113701.6250 - val_mse: 1113701.6250
Epoch 42/60
7711/7711 [==============================] - 9s 1ms/step - loss: 598612.6250 - mse: 598612.6250 - val_loss: 1137704.7500 - val_mse: 1137704.7500
Epoch 43/60
7711/7711 [==============================] - 11s 1ms/step - loss: 598125.8125 - mse: 598125.8125 - val_loss: 1116975.2500 - val_mse: 1116975.2500
Epoch 44/60
7711/7711 [==============================] - 10s 1ms/step - loss: 598072.0625 - mse: 598072.0625 - val_loss: 1121959.7500 - val_mse: 1121959.7500
Epoch 45/60
7711/7711 [==============================] - 12s 2ms/step - loss: 597850.9375 - mse: 597850.9375 - val_loss: 1108042.0000 - val_mse: 1108042.0000
Epoch 46/60
7711/7711 [==============================] - 9s 1ms/step - loss: 597401.3750 - mse: 597401.3750 - val_loss: 1112726.6250 - val_mse: 1112726.6250
Epoch 47/60
7711/7711 [==============================] - 9s 1ms/step - loss: 597065.1875 - mse: 597065.1875 - val_loss: 1136495.1250 - val_mse: 1136495.1250
Epoch 48/60
7711/7711 [==============================] - 10s 1ms/step - loss: 596339.3125 - mse: 596339.3125 - val_loss: 1125084.2500 - val_mse: 1125084.2500
Epoch 1/60
7711/7711 [==============================] - 11s 1ms/step - loss: 1132425.3750 - mse: 1132425.3750 - val_loss: 1509432.0000 - val_mse: 1509432.0000
Epoch 2/60
7711/7711 [==============================] - 9s 1ms/step - loss: 777156.0000 - mse: 777156.0000 - val_loss: 1482326.0000 - val_mse: 1482326.0000
Epoch 3/60
7711/7711 [==============================] - 9s 1ms/step - loss: 763858.1875 - mse: 763858.1875 - val_loss: 1458411.1250 - val_mse: 1458411.1250
Epoch 4/60
7711/7711 [==============================] - 9s 1ms/step - loss: 746270.7500 - mse: 746270.7500 - val_loss: 1415960.1250 - val_mse: 1415960.1250
Epoch 5/60
7711/7711 [==============================] - 11s 1ms/step - loss: 737213.4375 - mse: 737213.4375 - val_loss: 1446671.7500 - val_mse: 1446671.7500
Epoch 6/60
7711/7711 [==============================] - 11s 1ms/step - loss: 730479.3125 - mse: 730479.3125 - val_loss: 1400168.7500 - val_mse: 1400168.7500
Epoch 7/60
7711/7711 [==============================] - 9s 1ms/step - loss: 725888.9375 - mse: 725888.9375 - val_loss: 1417895.7500 - val_mse: 1417895.7500
Epoch 8/60
7711/7711 [==============================] - 9s 1ms/step - loss: 721182.6250 - mse: 721182.6250 - val_loss: 1378795.3750 - val_mse: 1378795.3750
Epoch 9/60
7711/7711 [==============================] - 9s 1ms/step - loss: 718407.6250 - mse: 718407.6250 - val_loss: 1426406.6250 - val_mse: 1426406.6250
Epoch 10/60
7711/7711 [==============================] - 9s 1ms/step - loss: 716383.8750 - mse: 716383.8750 - val_loss: 1435143.7500 - val_mse: 1435143.7500
Epoch 11/60
7711/7711 [==============================] - 9s 1ms/step - loss: 715198.0000 - mse: 715198.0000 - val_loss: 1336512.1250 - val_mse: 1336512.1250
Epoch 12/60
7711/7711 [==============================] - 9s 1ms/step - loss: 713517.8750 - mse: 713517.8750 - val_loss: 1429313.6250 - val_mse: 1429313.6250
Epoch 13/60
7711/7711 [==============================] - 9s 1ms/step - loss: 712042.6875 - mse: 712042.6875 - val_loss: 1345133.6250 - val_mse: 1345133.6250
Epoch 14/60
7711/7711 [==============================] - 9s 1ms/step - loss: 711390.2500 - mse: 711390.2500 - val_loss: 1342901.7500 - val_mse: 1342901.7500
Epoch 15/60
7711/7711 [==============================] - 9s 1ms/step - loss: 709819.3125 - mse: 709819.3125 - val_loss: 1386926.7500 - val_mse: 1386926.7500
Epoch 16/60
7711/7711 [==============================] - 9s 1ms/step - loss: 707437.2500 - mse: 707437.2500 - val_loss: 1405117.1250 - val_mse: 1405117.1250
Epoch 17/60
7711/7711 [==============================] - 9s 1ms/step - loss: 708080.6250 - mse: 708080.6250 - val_loss: 1342979.6250 - val_mse: 1342979.6250
Epoch 18/60
7711/7711 [==============================] - 9s 1ms/step - loss: 706751.9375 - mse: 706751.9375 - val_loss: 1344545.1250 - val_mse: 1344545.1250
Epoch 19/60
7711/7711 [==============================] - 9s 1ms/step - loss: 704926.8125 - mse: 704926.8125 - val_loss: 1361914.0000 - val_mse: 1361914.0000
Epoch 20/60
7711/7711 [==============================] - 9s 1ms/step - loss: 704361.1875 - mse: 704361.1875 - val_loss: 1417299.3750 - val_mse: 1417299.3750
Epoch 21/60
7711/7711 [==============================] - 9s 1ms/step - loss: 703948.7500 - mse: 703948.7500 - val_loss: 1435128.7500 - val_mse: 1435128.7500
Epoch 1/60
7711/7711 [==============================] - 9s 1ms/step - loss: 1181778.1250 - mse: 1181778.1250 - val_loss: 1731819.8750 - val_mse: 1731819.8750
Epoch 2/60
7711/7711 [==============================] - 9s 1ms/step - loss: 890525.5625 - mse: 890525.5625 - val_loss: 1731751.5000 - val_mse: 1731751.5000
Epoch 3/60
7711/7711 [==============================] - 9s 1ms/step - loss: 873378.3750 - mse: 873378.3750 - val_loss: 1655156.0000 - val_mse: 1655156.0000
Epoch 4/60
7711/7711 [==============================] - 9s 1ms/step - loss: 854325.0000 - mse: 854325.0000 - val_loss: 1652154.7500 - val_mse: 1652154.7500
Epoch 5/60
7711/7711 [==============================] - 9s 1ms/step - loss: 843888.6250 - mse: 843888.6250 - val_loss: 1695737.7500 - val_mse: 1695737.7500
Epoch 6/60
7711/7711 [==============================] - 9s 1ms/step - loss: 835854.7500 - mse: 835854.7500 - val_loss: 1600748.8750 - val_mse: 1600748.8750
Epoch 7/60
7711/7711 [==============================] - 9s 1ms/step - loss: 830025.0625 - mse: 830025.0625 - val_loss: 1592590.6250 - val_mse: 1592590.6250
Epoch 8/60
7711/7711 [==============================] - 9s 1ms/step - loss: 826837.4375 - mse: 826837.4375 - val_loss: 1589897.0000 - val_mse: 1589897.0000
Epoch 9/60
7711/7711 [==============================] - 9s 1ms/step - loss: 822391.6250 - mse: 822391.6250 - val_loss: 1566108.3750 - val_mse: 1566108.3750
Epoch 10/60
7711/7711 [==============================] - 9s 1ms/step - loss: 822212.3125 - mse: 822212.3125 - val_loss: 1548259.8750 - val_mse: 1548259.8750
Epoch 11/60
7711/7711 [==============================] - 9s 1ms/step - loss: 819643.1875 - mse: 819643.1875 - val_loss: 1571170.1250 - val_mse: 1571170.1250
Epoch 12/60
7711/7711 [==============================] - 9s 1ms/step - loss: 818968.4375 - mse: 818968.4375 - val_loss: 1558449.3750 - val_mse: 1558449.3750
Epoch 13/60
7711/7711 [==============================] - 9s 1ms/step - loss: 816988.0625 - mse: 816988.0625 - val_loss: 1564328.6250 - val_mse: 1564328.6250
Epoch 14/60
7711/7711 [==============================] - 9s 1ms/step - loss: 815708.3125 - mse: 815708.3125 - val_loss: 1606196.7500 - val_mse: 1606196.7500
Epoch 15/60
7711/7711 [==============================] - 9s 1ms/step - loss: 814335.6875 - mse: 814335.6875 - val_loss: 1560991.7500 - val_mse: 1560991.7500
Epoch 16/60
7711/7711 [==============================] - 9s 1ms/step - loss: 813446.6250 - mse: 813446.6250 - val_loss: 1599298.8750 - val_mse: 1599298.8750
Epoch 17/60
7711/7711 [==============================] - 9s 1ms/step - loss: 813120.2500 - mse: 813120.2500 - val_loss: 1525991.7500 - val_mse: 1525991.7500
Epoch 18/60
7711/7711 [==============================] - 9s 1ms/step - loss: 811167.9375 - mse: 811167.9375 - val_loss: 1526312.1250 - val_mse: 1526312.1250
Epoch 19/60
7711/7711 [==============================] - 9s 1ms/step - loss: 811342.2500 - mse: 811342.2500 - val_loss: 1542945.2500 - val_mse: 1542945.2500
Epoch 20/60
7711/7711 [==============================] - 9s 1ms/step - loss: 809676.8750 - mse: 809676.8750 - val_loss: 1578484.1250 - val_mse: 1578484.1250
Epoch 21/60
7711/7711 [==============================] - 9s 1ms/step - loss: 808908.6250 - mse: 808908.6250 - val_loss: 1570290.2500 - val_mse: 1570290.2500
Epoch 22/60
7711/7711 [==============================] - 9s 1ms/step - loss: 809047.8750 - mse: 809047.8750 - val_loss: 1523161.5000 - val_mse: 1523161.5000
Epoch 23/60
7711/7711 [==============================] - 9s 1ms/step - loss: 808180.5000 - mse: 808180.5000 - val_loss: 1560078.3750 - val_mse: 1560078.3750
Epoch 24/60
7711/7711 [==============================] - 9s 1ms/step - loss: 807311.0625 - mse: 807311.0625 - val_loss: 1545962.3750 - val_mse: 1545962.3750
Epoch 25/60
7711/7711 [==============================] - 10s 1ms/step - loss: 806318.3750 - mse: 806318.3750 - val_loss: 1525198.6250 - val_mse: 1525198.6250
Epoch 26/60
7711/7711 [==============================] - 9s 1ms/step - loss: 804609.8750 - mse: 804609.8750 - val_loss: 1529203.8750 - val_mse: 1529203.8750
Epoch 27/60
7711/7711 [==============================] - 11s 1ms/step - loss: 804607.8750 - mse: 804607.8750 - val_loss: 1570753.0000 - val_mse: 1570753.0000
Epoch 28/60
7711/7711 [==============================] - 9s 1ms/step - loss: 805101.6875 - mse: 805101.6875 - val_loss: 1542289.6250 - val_mse: 1542289.6250
Epoch 29/60
7711/7711 [==============================] - 9s 1ms/step - loss: 803180.8125 - mse: 803180.8125 - val_loss: 1524336.3750 - val_mse: 1524336.3750
Epoch 30/60
7711/7711 [==============================] - 9s 1ms/step - loss: 803533.3125 - mse: 803533.3125 - val_loss: 1539272.2500 - val_mse: 1539272.2500
Epoch 31/60
7711/7711 [==============================] - 9s 1ms/step - loss: 801790.0000 - mse: 801790.0000 - val_loss: 1530008.5000 - val_mse: 1530008.5000
Epoch 32/60
7711/7711 [==============================] - 9s 1ms/step - loss: 801666.6250 - mse: 801666.6250 - val_loss: 1505929.2500 - val_mse: 1505929.2500
Epoch 33/60
7711/7711 [==============================] - 10s 1ms/step - loss: 801340.0000 - mse: 801340.0000 - val_loss: 1545276.1250 - val_mse: 1545276.1250
Epoch 34/60
7711/7711 [==============================] - 9s 1ms/step - loss: 801959.5625 - mse: 801959.5625 - val_loss: 1515768.3750 - val_mse: 1515768.3750
Epoch 35/60
7711/7711 [==============================] - 9s 1ms/step - loss: 799388.7500 - mse: 799388.7500 - val_loss: 1519302.6250 - val_mse: 1519302.6250
Epoch 36/60
7711/7711 [==============================] - 9s 1ms/step - loss: 798700.5000 - mse: 798700.5000 - val_loss: 1518900.7500 - val_mse: 1518900.7500
Epoch 37/60
7711/7711 [==============================] - 9s 1ms/step - loss: 799982.1875 - mse: 799982.1875 - val_loss: 1556717.6250 - val_mse: 1556717.6250
Epoch 38/60
7711/7711 [==============================] - 9s 1ms/step - loss: 798784.0000 - mse: 798784.0000 - val_loss: 1516352.7500 - val_mse: 1516352.7500
Epoch 39/60
7711/7711 [==============================] - 11s 1ms/step - loss: 798727.1250 - mse: 798727.1250 - val_loss: 1519625.8750 - val_mse: 1519625.8750
Epoch 40/60
7711/7711 [==============================] - 9s 1ms/step - loss: 796833.0000 - mse: 796833.0000 - val_loss: 1528183.2500 - val_mse: 1528183.2500
Epoch 41/60
7711/7711 [==============================] - 9s 1ms/step - loss: 796501.0625 - mse: 796501.0625 - val_loss: 1512159.7500 - val_mse: 1512159.7500
Epoch 42/60
7711/7711 [==============================] - 9s 1ms/step - loss: 796180.5625 - mse: 796180.5625 - val_loss: 1521572.2500 - val_mse: 1521572.2500
Epoch 1/60
7711/7711 [==============================] - 9s 1ms/step - loss: 1278796.7500 - mse: 1278796.7500 - val_loss: 1949078.0000 - val_mse: 1949078.0000
Epoch 2/60
7711/7711 [==============================] - 9s 1ms/step - loss: 997129.2500 - mse: 997129.2500 - val_loss: 1925164.6250 - val_mse: 1925164.6250
Epoch 3/60
7711/7711 [==============================] - 10s 1ms/step - loss: 976509.5000 - mse: 976509.5000 - val_loss: 1883684.7500 - val_mse: 1883684.7500
Epoch 4/60
7711/7711 [==============================] - 9s 1ms/step - loss: 957808.1875 - mse: 957808.1875 - val_loss: 1901554.7500 - val_mse: 1901554.7500
Epoch 5/60
7711/7711 [==============================] - 9s 1ms/step - loss: 947281.0000 - mse: 947281.0000 - val_loss: 1812553.1250 - val_mse: 1812553.1250
Epoch 6/60
7711/7711 [==============================] - 9s 1ms/step - loss: 940146.9375 - mse: 940146.9375 - val_loss: 1846085.6250 - val_mse: 1846085.6250
Epoch 7/60
7711/7711 [==============================] - 9s 1ms/step - loss: 934947.1250 - mse: 934947.1250 - val_loss: 1797743.3750 - val_mse: 1797743.3750
Epoch 8/60
7711/7711 [==============================] - 9s 1ms/step - loss: 931173.2500 - mse: 931173.2500 - val_loss: 1767478.3750 - val_mse: 1767478.3750
Epoch 9/60
7711/7711 [==============================] - 9s 1ms/step - loss: 927560.5000 - mse: 927560.5000 - val_loss: 1772028.1250 - val_mse: 1772028.1250
Epoch 10/60
7711/7711 [==============================] - 9s 1ms/step - loss: 924397.8125 - mse: 924397.8125 - val_loss: 1786684.2500 - val_mse: 1786684.2500
Epoch 11/60
7711/7711 [==============================] - 9s 1ms/step - loss: 921408.6250 - mse: 921408.6250 - val_loss: 1819296.1250 - val_mse: 1819296.1250
Epoch 12/60
7711/7711 [==============================] - 9s 1ms/step - loss: 922285.3750 - mse: 922285.3750 - val_loss: 1752158.1250 - val_mse: 1752158.1250
Epoch 13/60
7711/7711 [==============================] - 9s 1ms/step - loss: 920026.8750 - mse: 920026.8750 - val_loss: 1828496.6250 - val_mse: 1828496.6250
Epoch 14/60
7711/7711 [==============================] - 9s 1ms/step - loss: 918136.6250 - mse: 918136.6250 - val_loss: 1757702.1250 - val_mse: 1757702.1250
Epoch 15/60
7711/7711 [==============================] - 9s 1ms/step - loss: 918691.1250 - mse: 918691.1250 - val_loss: 2002906.7500 - val_mse: 2002906.7500
Epoch 16/60
7711/7711 [==============================] - 9s 1ms/step - loss: 916749.6250 - mse: 916749.6250 - val_loss: 1817758.7500 - val_mse: 1817758.7500
Epoch 17/60
7711/7711 [==============================] - 9s 1ms/step - loss: 915257.8750 - mse: 915257.8750 - val_loss: 1736416.8750 - val_mse: 1736416.8750
Epoch 18/60
7711/7711 [==============================] - 9s 1ms/step - loss: 913891.8125 - mse: 913891.8125 - val_loss: 1744511.8750 - val_mse: 1744511.8750
Epoch 19/60
7711/7711 [==============================] - 9s 1ms/step - loss: 914595.6250 - mse: 914595.6250 - val_loss: 1753693.2500 - val_mse: 1753693.2500
Epoch 20/60
7711/7711 [==============================] - 9s 1ms/step - loss: 913603.1250 - mse: 913603.1250 - val_loss: 1728893.7500 - val_mse: 1728893.7500
Epoch 21/60
7711/7711 [==============================] - 9s 1ms/step - loss: 911602.8750 - mse: 911602.8750 - val_loss: 1747388.6250 - val_mse: 1747388.6250
Epoch 22/60
7711/7711 [==============================] - 9s 1ms/step - loss: 911774.9375 - mse: 911774.9375 - val_loss: 1732347.2500 - val_mse: 1732347.2500
Epoch 23/60
7711/7711 [==============================] - 9s 1ms/step - loss: 908607.0000 - mse: 908607.0000 - val_loss: 1724889.8750 - val_mse: 1724889.8750
Epoch 24/60
7711/7711 [==============================] - 9s 1ms/step - loss: 909152.6250 - mse: 909152.6250 - val_loss: 1732380.3750 - val_mse: 1732380.3750
Epoch 25/60
7711/7711 [==============================] - 9s 1ms/step - loss: 907809.1250 - mse: 907809.1250 - val_loss: 1727829.7500 - val_mse: 1727829.7500
Epoch 26/60
7711/7711 [==============================] - 9s 1ms/step - loss: 907681.1250 - mse: 907681.1250 - val_loss: 1730006.6250 - val_mse: 1730006.6250
Epoch 27/60
7711/7711 [==============================] - 9s 1ms/step - loss: 906352.7500 - mse: 906352.7500 - val_loss: 1726694.1250 - val_mse: 1726694.1250
Epoch 28/60
7711/7711 [==============================] - 9s 1ms/step - loss: 906139.6875 - mse: 906139.6875 - val_loss: 1758837.1250 - val_mse: 1758837.1250
Epoch 29/60
7711/7711 [==============================] - 9s 1ms/step - loss: 906636.4375 - mse: 906636.4375 - val_loss: 1764272.2500 - val_mse: 1764272.2500
Epoch 30/60
7711/7711 [==============================] - 9s 1ms/step - loss: 905355.1250 - mse: 905355.1250 - val_loss: 1843535.6250 - val_mse: 1843535.6250
Epoch 31/60
7711/7711 [==============================] - 9s 1ms/step - loss: 904173.2500 - mse: 904173.2500 - val_loss: 1750857.0000 - val_mse: 1750857.0000
Epoch 32/60
7711/7711 [==============================] - 9s 1ms/step - loss: 903505.5625 - mse: 903505.5625 - val_loss: 1775172.3750 - val_mse: 1775172.3750
Epoch 33/60
7711/7711 [==============================] - 9s 1ms/step - loss: 902576.0000 - mse: 902576.0000 - val_loss: 1708645.8750 - val_mse: 1708645.8750
Epoch 34/60
7711/7711 [==============================] - 9s 1ms/step - loss: 903073.8125 - mse: 903073.8125 - val_loss: 1741841.1250 - val_mse: 1741841.1250
Epoch 35/60
7711/7711 [==============================] - 9s 1ms/step - loss: 902317.1250 - mse: 902317.1250 - val_loss: 1734515.7500 - val_mse: 1734515.7500
Epoch 36/60
7711/7711 [==============================] - 9s 1ms/step - loss: 901961.6250 - mse: 901961.6250 - val_loss: 1773431.8750 - val_mse: 1773431.8750
Epoch 37/60
7711/7711 [==============================] - 9s 1ms/step - loss: 902059.1250 - mse: 902059.1250 - val_loss: 1752508.6250 - val_mse: 1752508.6250
Epoch 38/60
7711/7711 [==============================] - 9s 1ms/step - loss: 901389.4375 - mse: 901389.4375 - val_loss: 1710294.6250 - val_mse: 1710294.6250
Epoch 39/60
7711/7711 [==============================] - 9s 1ms/step - loss: 900894.8125 - mse: 900894.8125 - val_loss: 1745904.8750 - val_mse: 1745904.8750
Epoch 40/60
7711/7711 [==============================] - 10s 1ms/step - loss: 900231.1875 - mse: 900231.1875 - val_loss: 1752176.3750 - val_mse: 1752176.3750
Epoch 41/60
7711/7711 [==============================] - 9s 1ms/step - loss: 899684.0625 - mse: 899684.0625 - val_loss: 1726522.2500 - val_mse: 1726522.2500
Epoch 42/60
7711/7711 [==============================] - 9s 1ms/step - loss: 898448.3750 - mse: 898448.3750 - val_loss: 1714041.1250 - val_mse: 1714041.1250
Epoch 43/60
7711/7711 [==============================] - 9s 1ms/step - loss: 899241.1250 - mse: 899241.1250 - val_loss: 1724680.3750 - val_mse: 1724680.3750

# Metrics for each horizon for MLP:
resultados_mlp = []
path = "modelos_mlp_simples/{}".format(directory)

for target, i in enumerate(os.listdir(path)):
    path_temp = path + "/" + str(i)
    modelo_carregado = tf.keras.models.load_model(path_temp)
    mse, rmse, coef_p = metricas(X_test_std, y_test[:, target], {'mlp':modelo_carregado})
    resultados_mlp.append([mse, rmse, coef_p])

2409/2409 [==============================] - 2s 746us/step
Mean squared error: 247541.23993576627
Root Mean squared error: 497.5351645218318
Coef de pearson: 0.9902132723383218
###########################################

2409/2409 [==============================] - 2s 907us/step
Mean squared error: 368879.8296963908
Root Mean squared error: 607.354780747127
Coef de pearson: 0.9859220270338874
###########################################

2409/2409 [==============================] - 2s 815us/step
Mean squared error: 476382.2038747935
Root Mean squared error: 690.2044652672087
Coef de pearson: 0.9811503371802259
###########################################

2409/2409 [==============================] - 2s 855us/step
Mean squared error: 603839.6397278901
Root Mean squared error: 777.0711934745041
Coef de pearson: 0.976009306780267
###########################################

2409/2409 [==============================] - 2s 822us/step
Mean squared error: 748496.927420464
Root Mean squared error: 865.157169201333
Coef de pearson: 0.9701566298247793
###########################################

2409/2409 [==============================] - 2s 788us/step
Mean squared error: 911933.6948043354
Root Mean squared error: 954.9521950361366
Coef de pearson: 0.963925694795466
###########################################

2409/2409 [==============================] - 2s 904us/step
Mean squared error: 1086669.918192669
Root Mean squared error: 1042.4346109913413
Coef de pearson: 0.9563904985072705
###########################################

2409/2409 [==============================] - 2s 1ms/step
Mean squared error: 1302044.8054143423
Root Mean squared error: 1141.0717792559512
Coef de pearson: 0.9475641771071561
###########################################

2409/2409 [==============================] - 2s 914us/step
Mean squared error: 1466780.8947955202
Root Mean squared error: 1211.1073011073463
Coef de pearson: 0.9407577625656038
###########################################

2409/2409 [==============================] - 2s 961us/step
Mean squared error: 1660413.4333899617
Root Mean squared error: 1288.5703059553878
Coef de pearson: 0.9326020361695135
###########################################

4.2) Random Forest

# Training Random Forest models for each horizon:
resultados_rnd = []
for target in range(n_target):
    rnd_model = RandomForestRegressor(random_state=42)
    rnd_model.fit(X_train_std, y_train[:, target])
    print(f'Random Forest - {target}:')
    mse, rmse, coef_p = metricas(X_test_std, y_test[:, target], {'rnd':rnd_model})
    resultados_rnd.append([mse, rmse, coef_p])

4.3) XGBoost

# Training XGBoost models for each horizon:
resultados_xgb = []
for target in range(n_target):
    xgb_model = XGBRegressor(random_state=42)
    xgb_model.fit(X_train_std, y_train[:, target])

    print(f'XGBoost - {target}:')
    mse, rmse, coef_p = metricas(X_test_std, y_test[:, target], {'xgb':xgb_model})
    resultados_xgb.append([mse, rmse, coef_p])

4.4) Long Short term memory Neural Network (LSTM)

Long Short-Term Memory (LSTM) neural networks are a type of Recurrent Neural Network (RNN) that have the capability of remembering long-term information using an LSTM unit composed of three main "doors" that control the flow of information maintained or forgotten in the memory unit called the state cell. The name and functionality of each door can be seen below:

• Forget gate: It decides which information can be discarded from the state cell.

• Input gate: The input gate decides which new information should be stored in the state cell.

• Output gate: The output gate decides which information in the state cell should be used to generate the output.

This type of Recurrent Neural Network is great for tasks that involve long-term dependencies and contextualized information, making it suitable for solving time series problems, NLP problems, and more.

# Transforming the features into a 3D Matrix:
X_train_lstm = X_train_std.reshape(X_train_std.shape[0], 1, X_train_std.shape[1])
X_val_lstm = X_val_std.reshape(X_val_std.shape[0], 1, X_val_std.shape[1])

# LSTM's architecture:
def model_LSTM(device):
    with tf.device(device):
        model = tf.keras.Sequential()
        model.add(LSTM(1, batch_input_shape=(1, X_train_lstm.shape[1], X_train_lstm.shape[2]), stateful=True))
        model.add(tf.keras.layers.Dense(y_train.shape[1]))
        model.compile(loss='mean_squared_error', optimizer='adam')

    return model

# Training a LSTM for each horizon:
path = "modelos_lstm/{}".format(directory)
make_directory(path)

for target in range(n_target):
    
    # EarlyStopping callback:
    earlystopping = tf.keras.callbacks.EarlyStopping(patience=10, restore_best_weights=True)
    checkpoint = tf.keras.callbacks.ModelCheckpoint("{}/lstm_model{}.h5".format(path, target))
    modelo_lstm = model_LSTM('/device:GPU:0')

    for i in range(120):
        history = modelo_lstm.fit(X_train_lstm, y_train[:, target], epochs=1, batch_size=1, verbose=1, shuffle=False, 
        validation_data=(X_val_lstm, y_val[:, target]), validation_batch_size=1, callbacks=[earlystopping, checkpoint])
        modelo_lstm.reset_states()

# Metrics for the LSTM:
resultados_lstm = []
path = "modelos_lstm/{}".format(directory)

for target, i in enumerate(os.listdir(path)):
    path_temp = path + "/" + str(i)
    modelo_carregado = tf.keras.models.load_model(path_temp)
    
    print(f'LSTM - {target}:')
    mse, rmse, coef_p = metricas(X_test_std, y_test[:, target], {'LSTM':modelo_carregado})
    resultados_lstm.append([mse, rmse, coef_p])

4.5) Convolutional 1D Neural Network

The 1D Convolutional Neural Network is a type of Convolutional Neural Network that processes unidimensional data to extract information and patterns. This architecture is used to tackle time series problems, text classification and other problems, thanks to its capacity for preprocessing sequences of data, where the order is fundamentally important.

# Reshaping variables to fit as Input for a CNN 1D:
X_train_conv = np.array(X_train_std).reshape(X_train_std.shape[0], X_train_std.shape[1], 1)
X_val_conv = np.array(X_val_std).reshape(X_val_std.shape[0], X_val_std.shape[1], 1)
X_test_conv = np.array(X_test_std).reshape(X_test_std.shape[0], X_test_std.shape[1], 1)

# CNN 1D:
def timeseries_model_conv(device):
    with tf.device(device):
        model = tf.keras.Sequential()
        model.add(Conv1D(filters=64, kernel_size=7, activation="relu", padding="same", input_shape=(X_train_conv.shape[1], 1)))
        model.add(MaxPooling1D(pool_size=2))
        model.add(Conv1D(filters=128, kernel_size=3, activation="relu", padding="same"))
        model.add(MaxPooling1D(pool_size=2))
        model.add(Conv1D(filters=256, kernel_size=3, activation="relu", padding="same"))
        model.add(MaxPooling1D(pool_size=2))
        model.add(tf.keras.layers.Flatten())
        model.add(tf.keras.layers.Dense(128, activation="relu"))
        model.add(tf.keras.layers.Dropout(0.5))
        model.add(tf.keras.layers.Dense(64, activation="relu"))
        model.add(tf.keras.layers.Dense(1))
        model.compile(optimizer="adam", loss="mse", metrics=["mse"])
        return model

# Training a 1D CNN for each horizon:
path = "modelos_cnn/{}".format(directory)
make_directory(path)

for target in range(n_target):
    
    earlystopping = tf.keras.callbacks.EarlyStopping(patience=10, restore_best_weights=True)
    checkpoint = tf.keras.callbacks.ModelCheckpoint("{}/cnn_model{}.h5".format(path, target))
    model_conv = timeseries_model_conv("/device:GPU:0")

    history_conv = model_conv.fit(X_train_conv, y_train[:, target], validation_data=(X_val_conv, y_val[:, target]), 
                  epochs=120, callbacks=[earlystopping, checkpoint])

# Metrics fr the 1D CNN:
resultados_cnn = []
path = 'modelos_cnn/{}'.format(directory)

for index, model in enumerate(os.listdir(path)):
    path_temp = path + "/" + model
    modelo_carregado = tf.keras.models.load_model(path_temp)
    
    print(f'CNN 1D - {target}:')
    mse, rmse, coef_p = metricas(X_test_conv, y_test[:, target], {'CNN':modelo_carregado})
    resultados_cnn.append([mse, rmse, coef_p])

5) Results

# Plotting the result figures:
resultados_mlp = np.array(resultados_mlp)
resultados_rnd = np.array(resultados_rnd)
resultados_xgb = np.array(resultados_xgb)
resultados_lstm = np.array(resultados_lstm)
resultados_cnn = np.array(resultados_cnn)
metricas_iterar = ['MSE', 'RMSE', 'COEF_PEARSON']

# Creating a directory to store all the images:
path = 'imagens/{}'.format(directory)
make_directory(path)

for index, nome in enumerate(metricas_iterar):
    resultados_mlp_reshaped = resultados_mlp[:, index].reshape(-1, 1)
    resultados_rnd_reshaped = resultados_rnd[:, index].reshape(-1, 1)
    resultados_xgb_reshaped = resultados_xgb[:, index].reshape(-1, 1)
    resultados_lstm_reshaped = resultados_lstm[:, index].reshape(-1, 1)
    resultados_cnn_reshaped = resultados_cnn[:, index].reshape(-1, 1)

    array_resultados_mse = np.concatenate([resultados_mlp_reshaped, resultados_rnd_reshaped, 
                                          resultados_xgb_reshaped, resultados_lstm_reshaped,
                                          resultados_cnn_reshaped], axis=1)

    df_resultados = pd.DataFrame(array_resultados_mse, columns=['MLP', 'RANDOM_FOREST', 'XGBOOST', 'LSTM', 'CNN']).reset_index(names='Horizontes')
    df_resultados_melted = df_resultados.melt(id_vars='Horizontes', value_name=nome, var_name='Modelos')

    fig_resultado = px.line(df_resultados_melted, x='Horizontes', y=nome, hover_data=['Modelos'], color='Modelos',
    title='{} - 12 Horizontes'.format(nome))
    
    pl.io.write_image(fig=fig_resultado, file='imagens\{}\{}.jpg'.format(directory, nome), width=1000, height=500)