#!/usr/bin/env python
# coding: utf-8

# In[1]:


import numpy as np
import math as mth
import matplotlib.pyplot as plt
import sklearn.model_selection as ms
import sklearn.neural_network as nn
import sklearn.metrics as met
import scipy.stats as stt


# In[2]:


def F(x1, x2, x3, x4):
    y = 1.4 * x1 - 1.6 * x2 ** 2 + mth.sin(3 * x3) + 2.7 * abs(x4 - 1) + 2
    return y


# In[3]:


np.random.seed(50)
Min = -3
Max = +3
N = 4000
x = np.zeros((N, 4))
y = np.zeros((N))


# In[6]:


for i in range(0, N):
    x[i, 0] = np.random.uniform(Min, Max)
    x[i, 1] = np.random.uniform(Min, Max)
    x[i, 2] = np.random.uniform(Min, Max)
    x[i, 3] = np.random.uniform(Min, Max)
    y[i] = F(x[i, 0], x[i, 1], x[i, 2], x[i, 3])


# In[7]:


xtr, xte, ytr, yte = ms.train_test_split(x, y, train_size = 0.75, random_state = 45)


# In[8]:


NN = nn.MLPRegressor(hidden_layer_sizes = (40, 50),
                     activation = "relu",
                     solver = "adam",
                     learning_rate_init = 0.01,
                     max_iter = 5000,
                     random_state = 50)


# In[9]:


NN.fit(xtr, ytr.ravel())


# In[11]:


trPred = NN.predict(xtr)
tePred = NN.predict(xte)

trMSE = met.mean_squared_error(ytr, trPred)
teMSE = met.mean_squared_error(yte, tePred)

trPCC, _ = stt.pearsonr(ytr, trPred) 
tePCC, _ = stt.pearsonr(yte, tePred)

trMAPE = met.mean_absolute_percentage_error(ytr, trPred)
teMAPE = met.mean_absolute_percentage_error(ytr, trPred)


# In[12]:


print ("Train MSE : ", trMSE)
print ("Test MSE : ", teMSE)
print ("-" * 35)
print ("Train pcc : ", trPCC)
print ("Test pcc : ", tePCC)
print ("-" * 35)
print ("Train MAPE : ", trMAPE)
print ("Test MAPE : ", teMAPE)


# In[16]:


plt.figure(figsize = (7, 7))
plt.scatter(ytr[::2], trPred[::2], label = "Train (PCC = {})".format(round(trPCC, 4)), s = 1)
plt.scatter(yte[::2], tePred[::2], label = "Test (PCC = {})".format(round(tePCC, 4)), s = 1)
plt.plot([-15, +16], [-15, +16], label = "x = y", c = "r")
plt.title("MLP Regresstion Result Graph", fontsize = 18)
plt.xlabel("Target Values", fontsize = 12)
plt.ylabel("Predicted values", fontsize = 12)
plt.legend()


# In[ ]: