# Always suppress warnings in Jupyter notebooks
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))
 

# Quickly check basic info about each column
df.info()
 

# How many fraudulent versus normal transactions do we have?
print(df['Class'].value_counts())
 

# See some fraudulent cases
df[df['Class'] == 1].head(3)
 

# Explore column names and the first few more rows
print(df.columns.tolist())
df.head()
 

# Check for missing values: is anything blank?
print(df.isnull().sum().sum())
 

# Simple histogram plot of amounts by class
import seaborn as sns
import matplotlib.pyplot as plt
sns.histplot(df[df['Class'] == 0]['Amount'], bins=50, color='blue', label='Normal', alpha=0.5)
sns.histplot(df[df['Class'] == 1]['Amount'], bins=50, color='red', label='Fraudulent', alpha=0.7)
plt.legend()
plt.title('Transaction Amount By Fraud Status')
plt.xlabel('Amount')
plt.ylabel('Count')
plt.show()
 

# Preview how data looks over time
df['Hour'] = (df['Time'] // 3600) % 24
sns.countplot(x='Hour', hue='Class', data=df, palette={0:'blue',1:'red'})
plt.title('Fraud Counts By Hour of Day')
plt.ylabel('Transactions')
plt.xlabel('Hour')
plt.legend(['Normal','Fraud'], loc='upper right')
plt.show()
 

# Split data into inputs (X) and label (y)
X = df.drop(['Class'], axis=1)
y = df['Class']
 
# Split into train and test sets
from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.25, random_state=42, stratify=y)
print('Training set:', X_train.shape, y_train.shape)
print('Test set:', X_test.shape, y_test.shape)
 

# Let us build a simple model: Logistic Regression
from sklearn.linear_model import LogisticRegression
model = LogisticRegression(max_iter=1000, random_state=42, class_weight='balanced')
model.fit(X_train, y_train)
 

# Predict on test set and check accuracy
y_pred = model.predict(X_test)
from sklearn.metrics import accuracy_score, classification_report, confusion_matrix
print('Accuracy:', accuracy_score(y_test, y_pred))
print(classification_report(y_test, y_pred, target_names=['Not Fraud', 'Fraud']))
print(confusion_matrix(y_test, y_pred))
 

# Visualize confusion matrix as a heatmap
import seaborn as sns
cm = confusion_matrix(y_test, y_pred)
sns.heatmap(cm, annot=True, fmt='d', cmap='Reds', xticklabels=['Not Fraud','Fraud'], yticklabels=['Not Fraud','Fraud'])
plt.xlabel('Predicted')
plt.ylabel('Actual')
plt.title('Confusion Matrix')
plt.show()
 

# Try using a decision tree for comparison
from sklearn.tree import DecisionTreeClassifier
tree = DecisionTreeClassifier(max_depth=5, class_weight='balanced', random_state=42)
tree.fit(X_train, y_train)
y_tree_pred = tree.predict(X_test)
print('Accuracy (Decision Tree):', accuracy_score(y_test, y_tree_pred))
print(classification_report(y_test, y_tree_pred, target_names=['Not Fraud', 'Fraud']))
 

# See which features matter most for our decision tree
importances = tree.feature_importances_
features = X.columns
feat_importances = pd.Series(importances, index=features)
feat_importances.nlargest(10).plot(kind='barh')
plt.xlabel('Importance Score')
plt.title('Top 10 Feature Importances for Fraud Detection')
plt.show()
 

# Optional: try your own test amount and get the prediction
test_values = X_test.iloc[0].values.reshape(1, -1)
prediction = model.predict(test_values)[0]
if prediction == 1:
    result = "Fraud!"
else:
    result = "Not Fraud."
print("Prediction:", result)