ML - Lab 4 - ANN with training and testing error

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import sklearn
from math import floor

# Load the csv file
alldata = pd.read_csv("./alldata.txt")
N = len(alldata)
stop = floor(0.75*N)
xtrain = alldata.loc[0:stop-1, ["X1", "X2"]]
ytrain = alldata.loc[0:stop-1, "y"]
xtest = alldata.loc[stop:N, ["X1", "X2"]]
ytest = alldata.loc[stop:N, "y"]

# Scatter the training data
plt.scatter(xtrain[ytrain == 1].X1, xtrain[ytrain == 1].X2, c="blue", marker="o", label="1")
plt.scatter(xtrain[ytrain == 2].X1, xtrain[ytrain == 2].X2, c="red", marker="+", label="2")
plt.title("Initial points and their class")
plt.legend()
plt.show()


# ANN Creation
# 2 hidden layers, 20 hidden nodes/layer, max_iter = 10000
from sklearn.neural_network import MLPRegressor
clf = MLPRegressor(hidden_layer_sizes=(2, 2), tol=0.01)
clf = clf.fit(xtrain, ytrain)

# Training Error - Compare 'pred' with 'ytrain'
pred = clf.predict(xtrain)
trainingError = [(t - p) for (t, p) in zip(ytrain, pred)]
MA_TrE = np.mean(np.abs(trainingError))
print("Training Error = MA_TrE = " + str(MA_TrE))
plt.hist(trainingError, range=(-1, 1), rwidth=0.5)
plt.title("Training Error")
plt.show()

# Testing Error - Compare 'pred' with 'ytest'
pred = clf.predict(xtest)
testingError = [(t - p) for (t, p) in zip(ytest, pred)]
MA_TeE = np.mean(np.abs(testingError))
print("Testing Error = MA_TeE = " + str(MA_TeE))
plt.hist(testingError, range=(-1, 1), rwidth=0.5)
plt.title("Testing Error")
plt.show()

print(trainingError)
print(testingError)