Predictive model to identify fraudulent transactions in a synthetic financial dataset generated by PaySim. The pipeline followed for the project is as follows:
- Data Collection : Data obtained from here(6.3 million rows, 11columns)
- Data Pre Processing & Exploratory Data Analysis
- Data Visualization
- Handling of imbalanced data by Under Sampling
- Model building
- Model Evaluation
The model is built in an Anaconda Environment and Python 3.5.0. All the libraries which need to be downloaded are mentioned in requirements.txt.
- Install Python3 or above using Anaconda or any other method
- Install the requirements using pip install -r requirements.txt in Anaconda Prompt
- Clone this repo to your local machine
- Extract the zip file you downloaded
- Open a Python editor like Spyder or Jupyter notebook
- You may create and use a virtual environment to work on this project
- Set the working directory to the folder where you extracted the files
- Download the dataset from the link given in Data Collection above, name it as 'frauddata.csv' and save it in the working directory
- Run fraud.py