📊 Credit Card Fraud Detection Project

🎯 Objective

• The objective of this project is to develop a machine learning model that can predict whether a credit card transaction is fraudulent or not. 🕵️‍♂️💳

⚙️ Functionality

• Data analysis and cleaning 🧹
• Data exploration and visualization 🔍📊
• Training multiple machine learning models 🤖
• Evaluating and comparing models 📈
• Visualizing results and metrics 📉

🛠️ Tools Used

• Python 🐍
• Pandas for data manipulation 🐼
• NumPy for numerical operations 🔢
• Matplotlib and Seaborn for data visualization 📊🎨
• Scikit-learn for machine learning modeling and evaluation 🤖

🛤️ Development Process

1. Data Analysis and Cleaning 🧹

• Data Loading: Import the credit card transactions dataset 📥
• Basic Analysis: Explore the structure and basic statistics of the dataset 📊
• Duplicate Check: Identify and remove duplicate rows 🗑️
• Data Cleaning: Check for and handle missing values 🚫

2. Exploratory Data Analysis (EDA) 🔍

• Target Variable Distribution: Visualize the distribution of fraudulent and non-fraudulent transactions 📊
• Correlation Matrix: Analyze the correlation between variables 🔗
• Feature Distribution: Visualize the distribution of each feature 📈

3. Data Preprocessing 🧪

• Feature and Target Separation: Split the dataset into features (X) and target variable (y) ✂️
• Dataset Splitting: Divide the data into training and testing sets 🧩
• Feature Scaling: Normalize the features to improve model performance 📏

4. Machine Learning Modeling 🤖

• Logistic Regression: Train and evaluate a logistic regression model 📉
• Random Forest: Train and evaluate a Random Forest model 🌳
• Support Vector Machine (SVM): Train and evaluate an SVM model 🧠

5. Model Evaluation and Comparison 📈

• ROC Curve and AUC: Compare models using the ROC curve and area under the curve (AUC) 📊
• Precision-Recall Curve: Evaluate the precision and recall of the models 📉
• Additional Metrics: Calculate precision, recall, F1-score, and accuracy for each model 📏

📈 Results

• Logistic Regression: Good performance in precision but lower recall 📉
• Random Forest: Best balance between precision and recall 🌳
• SVM: High precision but lower recall compared to Random Forest 🧠

📊 Visualizations

• Target Variable Distribution: Bar chart showing the distribution of fraudulent and non-fraudulent transactions 📊
• Correlation Matrix: Heatmap showing the correlation between variables 🔗
• Feature Distribution: Histograms showing the distribution of each feature 📈
• ROC Curve: Plot comparing the ROC curves of the models 📉
• Precision-Recall Curve: Plot comparing the precision and recall of the models 📊

🗂️ Project Structure

• Notebook

📝 Conclusions

• Random Forest is the most balanced model for detecting credit card fraud 🌳
• Feature standardization and duplicate removal are crucial steps in data preprocessing 🧹
• Evaluating multiple metrics is essential for a comprehensive model comparison 📏

📬 Contact

• For any inquiries or collaborations, you can contact me at: jotaduranbon.com 📧