# import library
import pandas as pd, numpy as np, matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from sklearn.neural_network import MLPClassifier
from sklearn.metrics import accuracy_score, confusion_matrix, classification_report
from sklearn.datasets import make_moons, make_blobs, load_iris, load_wine, load_breast_cancer, load_digits

# =========================
# choose one dataset only
# =========================

# 1) moons
# X,y = make_moons(n_samples=200, noise=.2, random_state=42)
# df = pd.DataFrame(X, columns=["x1","x2"]); df["target"] = y

# 2) blobs
# X,y = make_blobs(n_samples=200, centers=3, random_state=42)
# df = pd.DataFrame(X, columns=["x1","x2"]); df["target"] = y

# 3) iris
d = load_iris()
X,y = d.data, d.target
df = pd.DataFrame(X, columns=d.feature_names); df["target"] = y

# 4) wine
# d = load_wine()
# X,y = d.data, d.target
# df = pd.DataFrame(X, columns=d.feature_names); df["target"] = y

# 5) breast cancer
# d = load_breast_cancer()
# X,y = d.data, d.target
# df = pd.DataFrame(X, columns=d.feature_names); df["target"] = y

# 6) digits
# d = load_digits()
# X,y = d.data, d.target
# df = pd.DataFrame(X); df["target"] = y

# EDA
print(df.head())
print(df.info())
print(df.describe())
print(df.isnull().sum())

# missing value treatment
for i in df.columns:
    df[i] = df[i].fillna(df[i].mean())

# normalization
f = lambda x: 0.1 + 0.8*(x-x.min())/(x.max()-x.min())
for i in df.columns[:-1]:
    df[i] = f(df[i])

# input-output
X = df.iloc[:,:-1]
y = df.iloc[:,-1]

# split
seed = 210412   # replace by your exam roll
x1,x2,y1,y2 = train_test_split(X,y,test_size=.2,random_state=seed)

# model fit
m = MLPClassifier(hidden_layer_sizes=(10,10), max_iter=2000, random_state=seed)
m.fit(x1,y1)

# prediction
p1,p2,p = m.predict(x1), m.predict(x2), m.predict(X)

# accuracy
print("Train Accuracy =", accuracy_score(y1,p1))
print("Test Accuracy  =", accuracy_score(y2,p2))
print("All Accuracy   =", accuracy_score(y,p))

# confusion matrix
print("\nConfusion Matrix (Test Data):")
print(confusion_matrix(y2,p2))

# classification report
print("\nClassification Report (Test Data):")
print(classification_report(y2,p2))

# plots
plt.scatter(range(len(y1)), y1, label="Actual Train", s=10)
plt.scatter(range(len(p1)), p1, label="Predicted Train", s=10)
plt.title("Train: Actual vs Predicted")
plt.legend()
plt.show()

plt.scatter(range(len(y2)), y2, label="Actual Test", s=10)
plt.scatter(range(len(p2)), p2, label="Predicted Test", s=10)
plt.title("Test: Actual vs Predicted")
plt.legend()
plt.show()

plt.scatter(range(len(y)), y, label="Actual All", s=10)
plt.scatter(range(len(p)), p, label="Predicted All", s=10)
plt.title("Overall: Actual vs Predicted")
plt.legend()
plt.show()