Skip to content

abdullahsaka/Capital_One-Data_Challenge

Repository files navigation

Fraud Detection Case Study

Objective:

Fraud detection is an important aspect of banking and financial companies. It’s essential for both financial institutions as well as their customers to be able to identify fraud quickly and accurately. My objective is to build a predictive model to determine whether a given transaction will be fraudulent or not. This will improve their business operations and provide better services to their customers.

Dataset:

Please download the dataset via the link : https://github.com/CapitalOneRecruiting/DS

Data Processing:

There are duplicated transactions in the dataset. One type of duplicated transaction is a reversed transaction, where a purchase is followed by a reversal. Another example is a multi-swipe, where a vendor accidentally charges a customer's card multiple times within a short time span.

ML algorithms:

Random Forest, Gradient Boosting and Extreme GradientBoosting (XGBoost)

Evaluation Criteria:

Accuracy, Precision Score and F-1 score


1.Exploratory Data Analysis - v1.0-EDA.ipynb

2.Data Processing and Preparation - v1.1-DataProcessing.ipynb

3.Modeling and Evaluation - v1.2-Modeling.ipynb

!! To execute v1.2-Modeling.ipynb, firstly you have to run v1.1-DataProcessing.ipynb and save processed data.