# Always suppress warnings for a beginner friendly experience
import warnings; warnings.filterwarnings("ignore")
import numpy as np
np.random.seed(42)
 

# Data setup
import pandas as pd
url = 'https://storage.googleapis.com/download.tensorflow.org/data/creditcard.csv'
df = pd.read_csv(url)
print(df.shape)
print(df.head(3))

# Let us check for missing values
print(df.isnull().sum().sum())

# Take a quick look at class distribution
print(df['Class'].value_counts())

# Plot class distribution
import matplotlib.pyplot as plt
import seaborn as sns
sns.countplot(x='Class', data=df, palette='Set2')
plt.title('Transaction type counts: 0 = normal, 1 = fraud')
plt.show()

# Summary stats for fraudulent and normal transactions
print(df.groupby('Class')['Amount'].describe())

# Drop the 'Time' column just to keep things simple
df = df.drop(['Time'], axis=1)

# Scale the 'Amount' column for fair modeling
from sklearn.preprocessing import StandardScaler
df['Amount_Scaled'] = StandardScaler().fit_transform(df[['Amount']])
df = df.drop(['Amount'], axis=1)

# Visualize feature distributions for fraud and normal
sns.kdeplot(df[df['Class']==0]['Amount_Scaled'], label='Normal', fill=True)
sns.kdeplot(df[df['Class']==1]['Amount_Scaled'], label='Fraud', fill=True)
plt.legend()
plt.title('Distribution of scaled amounts')
plt.show()

# Split the dataset into train and test sets (80 percent train, 20 percent test)
from sklearn.model_selection import train_test_split
X = df.drop('Class', axis=1)
y = df['Class']
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42, stratify=y)
print(X_train.shape, X_test.shape)

# Train a Random Forest Classifier on training data
from sklearn.ensemble import RandomForestClassifier
rf = RandomForestClassifier(n_estimators=100, random_state=42, n_jobs=-1)
rf.fit(X_train, y_train)

# Predict labels for your test set
y_pred = rf.predict(X_test)

# Evaluate results: accuracy, recall, and confusion matrix
from sklearn.metrics import classification_report, confusion_matrix
print(classification_report(y_test, y_pred, digits=4))
print(confusion_matrix(y_test, y_pred))

# Visualize confusion matrix for clearer understanding
import numpy as np
cm = confusion_matrix(y_test, y_pred)
sns.heatmap(cm, annot=True, fmt='d', cmap='Blues', xticklabels=['Normal','Fraud'], yticklabels=['Normal','Fraud'])
plt.xlabel('Predicted')
plt.ylabel('Actual')
plt.title('Confusion matrix')
plt.show()

# See the top features that help detect fraud
importances = rf.feature_importances_
indices = np.argsort(importances)[-10:][::-1]
plt.figure(figsize=(8,4))
sns.barplot(x=importances[indices], y=X.columns[indices], orient='h')
plt.title('Top 10 important features in fraud detection')
plt.show()

# Try your model on a custom made up transaction
sample = X_test.sample(1, random_state=1)
prediction = rf.predict(sample)
print('Fraudulent' if prediction[0] == 1 else 'Normal')

# Interactive prediction: enter your own transaction details
inputs = []
for col in X.columns:
    val = float(input(f'Enter value for {col}: '))
    inputs.append(val)
custom_pred = rf.predict([inputs])
print('Fraudulent' if custom_pred[0]==1 else 'Normal')