MLP Regressão

import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
from sklearn.metrics import mean_squared_error

entradaDadosTreino = np.sort(np.random.uniform(-2*np.pi, 2*np.pi, 1000))[:, np.newaxis]
saidaDadosTreino = (np.sin(entradaDadosTreino)/entradaDadosTreino).ravel()

entradaDadosTest = np.arange(-2*np.pi, 2*np.pi, 0.001)[:, np.newaxis]

model = tf.keras.models.Sequential([
  tf.keras.Input(shape=(1,)),
  tf.keras.layers.Dense(500, activation= "relu"),
  tf.keras.layers.Dense(100, activation= "relu"),
  tf.keras.layers.Dense(50, activation= "relu"),
  tf.keras.layers.Dense(1)
])
model.compile(loss= "mean_squared_error" ,
              optimizer="adam",
              metrics=["mean_squared_error"])

returns = model.fit(entradaDadosTreino, saidaDadosTreino, epochs=100)



preditTest = model.predict(entradaDadosTest)

predTrain= model.predict(entradaDadosTreino)
print("RMSE Treino = ",np.sqrt(mean_squared_error(saidaDadosTreino,predTrain)))

saidaDadosTest = (np.sin(entradaDadosTest)/entradaDadosTest).ravel()
print("RMSE Test = ",np.sqrt(mean_squared_error(saidaDadosTest,preditTest)))





fig = plt.subplots(2)
plt.subplots_adjust(hspace = 1.0)
plt.subplot(2, 1, 1)
plt.plot(entradaDadosTreino,saidaDadosTreino)
plt.xlabel('X')
plt.ylabel('Y')
plt.xlim(-2*np.pi, 2*np.pi)
plt.title(r'$Sinc(x) = \frac{\mathrm{sin}(x)}{x}$', fontsize=20)
possicaoLegendaX= [-2*np.pi , -np.pi,0, np.pi , 2*np.pi]
legendaX = ['$2\pi$','$-\pi$','0', '$\pi$', '$2\pi$']
plt.xticks(possicaoLegendaX, legendaX)
plt.subplot(2, 1, 2)
plt.plot(entradaDadosTest,preditTest,'r')
plt.xlabel('X')
plt.ylabel('Y')
plt.xlim(-2*np.pi, 2*np.pi)
plt.title(r'$Predict \approx Sinc(x)$', fontsize=20)
possicaoLegendaX= [-2*np.pi , -np.pi,0, np.pi , 2*np.pi]
legendaX = ['$2\pi$','$-\pi$','0', '$\pi$', '$2\pi$']
plt.xticks(possicaoLegendaX, legendaX)
plt.show()