from http.client import HTTPResponse
from urllib.request import urlopen
import pandas
from sklearn.model_selection import train_test_split
from sklearn.tree import DecisionTreeClassifier, plot_tree
from sklearn.linear_model import LogisticRegression
from sklearn.neural_network import MLPClassifier
from sklearn.metrics import confusion_matrix, ConfusionMatrixDisplay, davies_bouldin_score, roc_curve
from sklearn.cluster import KMeans
from sklearn.decomposition import PCA
import matplotlib.pyplot as plt

# 1. feladat

data = urlopen('https://arato.inf.unideb.hu/ispany.marton/MachineLearning/Datasets/banknote_authentication.txt')
lines = []
osztalyok = set()
for line in data.read().decode('utf-8').split('\r\n'):
    lines.append(line.split(','))
    osztalyok.add(line.split(',')[-1])

print('rekordok: ', len(lines))
print('attributumok:', len(lines[0]))
print('osztalyok:', len(osztalyok))

# 2. feladat

df = pandas.DataFrame(lines, columns=['variance', 'skewness', 'curtosis', 'entropy', 'target'])

# pandas.plotting.andrews_curves(df, 'target', color=['blue', 'red'])

# 3. feladat

X_train, X_test, y_train, y_test = train_test_split(df.iloc[:, :4], df['target'],
                                                    test_size=0.2, shuffle=True, random_state=100)

# 4. feladat

scores = dict()

dec_tree = DecisionTreeClassifier(max_depth=5, criterion='entropy')
dec_tree.fit(X_train, y_train)
dec_tree_score = dec_tree.score(X_test, y_test)
scores['dec_tree'] = dec_tree_score
print(dec_tree_score)

log_reg = LogisticRegression(solver='liblinear')
log_reg.fit(X_train, y_train)
log_reg_score = log_reg.score(X_test, y_test)
scores['log_reg'] = log_reg_score
print(log_reg_score)

neural = MLPClassifier(hidden_layer_sizes=(2), activation='logistic')
neural.fit(X_train, y_train)
neural_score = neural.score(X_test, y_test)
scores['neural'] = neural_score
print(neural_score)

for k, v in scores.items():
    if v == max(scores.values()):
        print('legjobb:', k, ': ', v)

# 5. feladat

plot_tree(dec_tree)
plt.show()

y_pred = dec_tree.predict(X_test)
cm = confusion_matrix(y_test, y_pred)
x = ConfusionMatrixDisplay(cm)
x.plot()
plt.show()

# 6. feladat

roc_curve()

# 7. feladat

kmeans2 = KMeans(n_clusters=2)
kmeans2.fit(df.iloc[:, :4], df['target'])
kmeans11 = KMeans(n_clusters=11)
kmeans11.fit(df.iloc[:, :4], df['target'])
db2 = davies_bouldin_score(df.iloc[:, :4], kmeans2.labels_)
db11 = davies_bouldin_score(df.iloc[:, :4], kmeans11.labels_)
print(db2, db11)