Fraud-Detection-by-Machine-Learning

Fraud Detection Model

This code reads in a dataset of financial transactions and builds a model to predict whether a given transaction is fraudulent or not.

Dataset

The dataset used in this code is fraud.csv. It contains the following columns:

• step: This is a numerical column representing the hour in the day (0 to 23) in which the transaction occurred.
• type: This is a categorical column representing the type of transaction (e.g. payment, transfer, cash-out).
• amount: This is a numerical column representing the amount of the transaction.
• nameOrig: This is a categorical column representing the customer who initiated the transaction.
• oldbalanceOrg: This is a numerical column representing the customer's balance before the transaction.
• newbalanceOrig: This is a numerical column representing the customer's balance after the transaction.
• nameDest: This is a categorical column representing the customer who is the recipient of the transaction.
• oldbalanceDest: This is a numerical column representing the recipient's balance before the transaction.
• newbalanceDest: This is a numerical column representing the recipient's balance after the transaction.
• isFraud: This is a binary column representing whether the transaction was fraudulent (1) or not (0).
• isFlaggedFraud: This is a binary column representing whether the transaction was flagged as fraudulent (1) or not (0).

Model

The code uses a Linear Regression model from the sklearn library to predict whether a given transaction is fraudulent or not. The dataset is split into training and testing sets using the train_test_split function, and the model is fit on the training data. The accuracy of the model is then evaluated on the testing data.