hariprasath-v / Hackerearth_transunion-data-science-analytics-hiring-challenge_2022

Machine learning model to classify the credit score based on people bank details and credit related information.

Hackerearth_transunion-data-science-analytics-hiring-challenge_2022

Competition hosted on Hackerearth

About

Machine learning model to classify the credit score based on people bank details and credit related information.

Final Score 78.03

Evaluation Metric is roc_wighted_ovr

File information

• HE_transunion_data_science_analytics_hiring_challenge_EDA.ipynb

  Packages Used,

   * seaborn
   * Pandas
   * klib
   * Numpy
   * Matplotlib
   * re
  

  Basic Exploratory Data Analysis

• transunion-data-science-analytics-hiring-challenge_model.ipynb

  Packages Used,

    * Sklearn
    * re
    * Pandas
    * Numpy
    * Matplotlib
    * catboost
  

  Data Pre-processing

  Created catboost classifier with 5-fold stratified cross validatation and tuned the hyperparameters with optuna framework.

  Model explanation with SHAP

  More information about data pre-processing and model tuning

Catboost classifier model feature importances

Top features impact for Good class

Top features impact for Standard class

Top features impact for Poor class