Creditcard Fraud Detection

The project focused on classifying fraud transactions from normal transactions using data with 284,807 observations against 30 features. After exploratory data analysis and preprocessing we noticed that the distribution of the two classes is biased, thus, imbalanced in favor of the normal transactions, thus, fraud transaction is less than 0.2% of the total observations. We therefore employed various techniques necessary to make meaning to the classification algorithms and performances using the Support Vector Machine, Random Forest Classifier and Artificial Neural Network models. The objective of the project to grants the skills in handling imbalanced dataset for machine learning problems. This article shall be of great help to battling imbalanced datasets. Check out notebook here

Recommended credit: I. K. Anni, Kobby. (2021). “Deep learning mini project.” Machine & deep learning projects. 1(3).