numpy pandas for importing dataset matplotlib for data visualization seaborn for plotting heat map sklearn for ml algorithm
Dataset of around 284807 credit card transactions. Highly imbalanced with around 0.001 class imbalance.
Seperated the dataset to x & y. x contains around 30 parameters which are a result of PCA dimensionality reduction. y represent the classes of each of the datapoints. 0 means normal transaction 1 means fraudulent.
Plotting the correlation matrix to find if there are any relation between the parameters
Performed a basic downsampling to improve class imbalance.
Isolation forest & Local outlier factor.
Second algorithm worked better than the first. Support vector machine wasn't used because it would take longer to train.
Achieved precision of 0.78 and recall of 0.67 on the imbalanced class.
