TRANSACTION FRAUD

Project

This is a project for the subject Machine Learning of CUNEF Master´s in Data Science. The objective of this research is to obtain the probability of fraud in different transactions. It will be applied different types of Machine Learning models comparing the results that were found and determine the model that is the optimum.

What did we use?

• Python 3.9.13
• Visual Studio Code
• Jupyter Notebook

Index

0. EDA
1. Preprocessing
2. Models:
   • 2.1. Base model (Dummy Model)
   • 2.2. GLM Ridge
   • 2.3. Logistic Regression with Lasso
   • 2.4. Random Forest
   • 2.5. Light GBM
   • 2.6. Gradient Boosting Classifier
   • 2.7. Support Vector Machine
   • 2.8. XGBoost
3. Model Selection
4. Interpretability

Content of the repository

• data:

  • Raw: Documents downloaded from the source of the dataset.
  • Processed data: Data dictionay processed.

• images: Pictures used in the differents notebooks.

• notebooks: Notebooks of the project.

• models: Pickles of the different models.

• env: Requirements of the environment

Authors

Victor Viloria Vázquez

• Email: victor.viloria@cunef.edu
• Linkedin: https://www.linkedin.com/in/vicviloria/

Antonio Nogués Podadera:

• Email: antonio.nogues@cunef.edu
• Linkedin: https://www.linkedin.com/in/antonio-nogu%C3%A9s-podadera/

Project Link:

https://github.com/ComputingVictor/Payments_Fraud_Practice