Fraud Detection - Team Naive Baes

Question Statement

25Credit is a credit card company which requires a model to detect fraud cases quickly and accurately. Not only do they want a model, but also a business solution on implementing the model in their system. You can find the detailed business case in the documents folder above.

Methodology

1. Exploratory Data Analysis

• Summary Statistics
• Duplicate Values Inspection
• Missing Values Inspection
• Distribution Plot Analysis
• Fraud Class Analysis
• Outlier Inspection

2. Data Pre-Processing

• Normalisation
• Missing Value Imputation
• Feature Selection
• Data Resampling
• Splitting the Dataset

3. Model Building

• Logistic Regression
• Decision Trees
• Random Forest
• Random Forest + AdaBoost

4. Evaluation Metrics

• F1 Score
• Precision
• Recall
• Weighted F1 Score
• ROC Area Under Curve

5. Model Evaluation

• Learning Curve
• K-fold Cross Validation

Business Solution

Equally important as our model was our business solution and our recommendation for the company to implement our model into their existing system. You can find the recommendation in the slide deck in the documents folder above as well

Overall Rank

1st Runners up :)

Contributors:

Muskaan Gupta, Shubham Periwal, Zhuo Yunying (Kaelyn)