Skip to content

gsorry/datathon-2019

Repository files navigation

Raiffeisen Recommender System

Bank dataset is not a provided here because it is too big. Need to be extracted in this folder: 'Datathon_sample_final1.csv'

Clean Data

The first step is to clean the raw data from dataset. Therefore, the notebook 'RFB - Data Cleaning.ipynb' should be started first. It will generate a new dataset: 'RFB - Clean Data.csv'

Load External Data

Another step is to load external data obtained from The Statistical Office of the Republic of Serbia and Open data portal of Republic of Serbia:

  • 'RFB - Potrosacke Cene Indeksi.txt'
  • 'RFB - Strukturna Imovina.txt'

and Synthetic label from:

  • 'RFB - Prvi Zajam.txt'

NOTE: Change txt extension to csv (due to version controll isues).

It will generate a new dataset: 'RFB - Merged Data.csv'

EDA

A simple EDA is described in Notebook 'RFB - EDA.ipynb'.

Machine Learning Algorithms

To run, train, and test ML models, run the following notebooks:

  • RFB - Classification Random Forest.ipynb
  • RFB - Classification Random LightGBM.ipynb

About

Datathon usecase

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published