The goal of this project is to identify instances of blockchain fraud through the application of machine learning. To do so, a dataset with various features such as transaction values, time intervals between transactions, etc., is utilized along with a binary label (FLAG) indicating whether a fraud has occurred (1) or not (0). The project will follow a supervised learning approach, where several classification models will be evaluated and one or more will be chosen based on their balance of recall score and false positive predictions.
- Pandas:
Run the following in terminal: pip install pandas
More info: https://pandas.pydata.org/pandas-docs/stable/getting_started/install.html
- Numpy:
Run the following in terminal: pip install numpy
More info: https://numpy.org/install/
- Sklearn: Run the following in terminal: pip install -U scikit-learn
More info: https://scikit-learn.org/stable/install.html
- Seaborn:
Run the following in terminal: pip install seaborn
More info: https://seaborn.pydata.org/installing.html
- Matplotlib:
Run the following in terminal: python -m pip install -U matplotlib
More info: https://matplotlib.org/stable/users/installing.html
-
Open the .ipynb file in Jupyter Lab
-
Ensure the dataset (transaction_dataset.csv file) submitted as part of the project is in the current directory.
-
Run all cells from the top