# Always start by suppressing warnings for a cleaner notebook.
import warnings
import numpy as np
np.random.seed(42)
warnings.filterwarnings("ignore")
 

# Data setup: Breast Cancer Dataset from scikit learn
from sklearn.datasets import load_breast_cancer
data = load_breast_cancer()
X, y = data.data, data.target
print("Feature matrix shape:", X.shape)
print("Target label shape:", y.shape)
print("First five labels:", y[:5])
print("First five samples:")
print(X[:5])

# The target labels are 0 for malignant and 1 for benign.
print("Target names:", data.target_names)
print("Feature names:", data.feature_names)
print("Number of samples:", len(y))

# Let us split our dataset: 80% for training, 20% for testing.
from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.2, random_state=42
)
print("Training samples:", X_train.shape[0])
print("Testing samples:", X_test.shape[0])

# Data check: See if the classes are balanced in both sets.
import numpy as np
print("Training set label counts:", np.bincount(y_train))
print("Testing set label counts:", np.bincount(y_test))

# Optional: Standardize features so all measurements are on a similar scale.
from sklearn.preprocessing import StandardScaler
scaler = StandardScaler()
X_train_scaled = scaler.fit_transform(X_train)
X_test_scaled = scaler.transform(X_test)

# Train a Logistic Regression classifier.
from sklearn.linear_model import LogisticRegression
clf = LogisticRegression(max_iter=10000, random_state=42)
clf.fit(X_train_scaled, y_train)
 

# Check the accuracy of the model on training and test data.
train_accuracy = clf.score(X_train_scaled, y_train)
test_accuracy = clf.score(X_test_scaled, y_test)
print(f"Training accuracy: {train_accuracy:.2f}")
print(f"Test accuracy: {test_accuracy:.2f}")

# Look deeper: Show a confusion matrix and a classification report.
from sklearn.metrics import confusion_matrix, classification_report
y_pred = clf.predict(X_test_scaled)
cm = confusion_matrix(y_test, y_pred)
cr = classification_report(y_test, y_pred, target_names=data.target_names)
print("Confusion Matrix:\n", cm)
print("\nClassification Report:\n", cr)

# Visualize: Plot a few features to understand separation.
import matplotlib.pyplot as plt
plt.figure(figsize=(7,5))
plt.scatter(X_train_scaled[:, 0], X_train_scaled[:, 1], c=y_train, cmap='coolwarm', alpha=0.6)
plt.xlabel(data.feature_names[0])
plt.ylabel(data.feature_names[1])
plt.title("Cancer cell samples (first two features)")
plt.show()

# Train and test a Support Vector Machine (SVM) classifier.
from sklearn.svm import SVC
svm_model = SVC(kernel='linear', random_state=42)
svm_model.fit(X_train_scaled, y_train)
svm_test_acc = svm_model.score(X_test_scaled, y_test)
print(f"SVM test accuracy: {svm_test_acc:.2f}")

# Bonus: Feature importance with logistic regression coefficients.
import pandas as pd
coef = clf.coef_[0]
importance = pd.Series(coef, index=data.feature_names)
print(importance.sort_values(key=abs, ascending=False).head(10))

# Make a prediction for a single sample (from the test set).
sample = X_test_scaled[0]
true_label = y_test[0]
predicted_label = clf.predict([sample])[0]
print(f"True label: {data.target_names[true_label]}")
print(f"Predicted label: {data.target_names[predicted_label]}")

# You try: Predict for a new (simulated) sample.
import numpy as np
user_sample = np.mean(X_train, axis=0) + np.random.randn(X_train.shape[1]) * 0.1
user_sample_scaled = scaler.transform([user_sample])
prediction = clf.predict(user_sample_scaled)[0]
print(f"Predicted label for your made up sample: {data.target_names[prediction]}")

# Mini project: Ask the user for measurements and predict cancer cell type.
user_features = []
for name in data.feature_names[:5]:
    val = float(input(f"Enter value for {name}: "))
    user_features.append(val)
user_features = user_features + list(np.mean(X_train, axis=0)[5:])
user_features_scaled = scaler.transform([user_features])
prediction = clf.predict(user_features_scaled)[0]
print(f"Prediction: {data.target_names[prediction]}")