Credit Card Fraud Detection Using Machine Learning Project
Are you visiting from India?
For the best experience and local offerings, please visit our India-specific store:
š https://vtupulse.com
The Project will be emailed immediately upon purchaseĀ
Following are the contents of the project
- Source Code
- Project Report (Both PDF and Word files)Ā
- Instructions to install the necessary Software and Libraries
- Step-by-step instructions to execute the project
Project Description
The project aims to predict fraudulent credit card transactions using machine learning models. This is crucial from the bankās and the customerās perspectives. The banks cannot afford to lose their customersā money to fraudsters. Every fraud is a loss to the bank as the bank is responsible for the fraudulent transactions.
The dataset contains credit card transactions made over a period of two days by European cardholders in September 2013. This dataset presents transactions that occurred in two days, with 492 frauds out of 284,807 transactions. The dataset is highly unbalanced; the positive class (frauds) accounts for 0.172% of all transactions. We need to take care of the data imbalance while building the model and come up with the best model by trying various algorithms.
Video Demonstration of Credit Card Fraud Detection Using Machine Learning Project
Ā
Steps in Implementing Credit Card Fraud Detection:-
The steps are broadly divided into the below steps. The sub-steps are also listed while we approach each of the steps.
- Reading, understanding, and visualizing the data
- Preparing the data for modeling
- Building the model
- Evaluate the model
We have used a total of 5 algorithms in our project
- Logistic Regression
- K-Nearest Neighbours
- Decision Tree
- Random Forest
- xgboost
Six Techniques are used for undersampling or oversampling
- Random oversampling
- SMOTE Oversampling
- Random Undersampling
- Tomek Links Undersampling
- Cluster centroids undersampling
- SMOTE + Tomek Links
Ā
We have compared each of the 5 algorithms in all 7 (Normal data + Six undersampled or oversampled data) scenarios comparing their accuracy through the Area under the curve of the ROC Curve.
Ā
The results showed that, in the case of fraud detection, the performance of a classifier can be significantly improved when sampling methods are used to rebalance the two classes.
Ā
Given the large imbalance ratio and a number of transactions, over-sampling should perhaps be favored with respect to under-sampling techniques. We have analyzed different types of oversampling and undersampling techniques. The various types were Random oversampling and SMOTE technique for oversampling and random, Tomek links and cluster centroids for undersampling. We also used a combination of SMOTE and Tomek links. We found out that SMOTE and SMOTE with Tomek Links gave very good results as compared to others. Cluster centroids performed the worst.
Ā
Installation of required software and libraries
1. Install the Anaconda Python Package
Ā
2. Extract the downloaded project into your machine
Ā
3. Follow the link to Download the dataset and Extract it into the project directory
Ā
4. Open the anaconda prompt (It will open in the default user directory)
Ā
5. Use the cd command to change the directory to the project folder
Ā
Example:
>> cd Path_of_project_folder
Ā
2. Install the requirements.txt using the following command
Ā
>>pip install -r requirements.txt
Ā
Follow the steps to run the project
Ā
1. Open the anaconda prompt (It will open in the default user directory)
Ā
2. Use the cd command to change the directory to the project folder
Ā
Example:
>> cd Path_of_project_folder
Ā
3. Open Jupyter Notebook
Ā
>>jupyter notebook
Ā
4. Open the Credit Card Fraud Detection Using Machine Learning - Local Version
Ā
5. Execute all cells
Ā
Happy Learning
Still need help to set up and execute the project
- Setup and modification are paid services based on requirements.
