hariprasath-v / Zindi_UmojaHack-India-Income-Prediction-Challenge

Create a machine learning model to predict whether an individual earns above 50,000 in a specific currency or not.

Zindi_UmojaHack-India-Income-Prediction-Challenge

Competition hosted on zindi.africa

About

Create a machine learning model to predict whether an individual earns above 50,000 in a specific currency or not.

Final Score is 0.613205338

Leaderboard Rank is 30

Evaluation Metric is F1 score.

File information

• zindi-income-prediction-challenge-umojahack-eda.ipynb

  Basic Exploratory Data Analysis

  Packages Used,

   * seaborn
   * Pandas
   * Numpy
   * Matplotlib
  

• zindi-income-prediction-challenge-umojahack-model.ipynb

  Data Pre-processing and model.

  Packages Used,

    * Sklearn
    * Pandas
    * Numpy
    * Matplotlib
    * catboost    
  

  Created catboost classifier model and evaluate the validation data with f1 score.

  For more detailed information about the model.

Catboost model optimal threshold(0.3862)

Based on the optimal threshold,

Train data classification report,

Validation data classification report,

Catboost – SHAP feature importances

Catboost – SHAP top feature impact

SHAP Feature impact for single observation(class 0)

SHAP Feature impact for single observation(class 1)