train_model.py

import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import classification_report, confusion_matrix, precision_recall_curve, auc
from imblearn.over_sampling import SMOTE
import joblib
import os

# Load data
print("Loading data...")
df = pd.read_csv('c:/card/creditcard.csv')

# Preprocessing
print("Preprocessing...")
scaler_amount = StandardScaler()
scaler_time = StandardScaler()
df['scaled_amount'] = scaler_amount.fit_transform(df['Amount'].values.reshape(-1, 1))
df['scaled_time'] = scaler_time.fit_transform(df['Time'].values.reshape(-1, 1))

# Drop original Time and Amount
df.drop(['Time', 'Amount'], axis=1, inplace=True)

# Define X and y
X = df.drop('Class', axis=1)
y = df['Class']

# Split data
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42, stratify=y)

# Handle imbalance with SMOTE
print("Applying SMOTE to balance training data...")
sm = SMOTE(random_state=42)
X_train_res, y_train_res = sm.fit_resample(X_train, y_train)

print(f"Original training shape: {X_train.shape}")
print(f"Resampled training shape: {X_train_res.shape}")

# Train Model
print("Training Random Forest Classifier (this might take a minute)...")
model = RandomForestClassifier(n_estimators=50, max_depth=10, random_state=42, n_jobs=-1)
model.fit(X_train_res, y_train_res)

# Evaluate
print("Evaluating model...")
y_pred = model.predict(X_test)
print("\nConfusion Matrix:")
print(confusion_matrix(y_test, y_pred))
print("\nClassification Report:")
print(classification_report(y_test, y_pred))

# Save model and scalers
print("Saving model and scalers...")
joblib.dump(model, 'c:/card/fraud_model.joblib')
joblib.dump(scaler_amount, 'c:/card/scaler_amount.joblib')
joblib.dump(scaler_time, 'c:/card/scaler_time.joblib')

print("Done! Files saved to c:/card/")