hariprasath-v / Hackerearth_tredence-data-scientist-hiring-challenge

Machine Learning model to predict the weekly dispatch count of the warehouse.

Hackerearth_tredence-data-scientist-hiring-challenge

Competition hosted on Hackerearth

About

Machine Learning model to predict the weekly dispatch count of the warehouse based on the different warehouses' product-wise daily dispatch data.

Final Score 82.4

Evaluation Metric is Mean Absolute Percentage Error.

File information

• tredence-data-scientist-hiring-challenge_EDA.ipynb

  Packages Used,

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

  Basic Exploratory Data Analysis

• tredence-data-scientist-hiring-challenge_Model.ipynb

  Packages Used,

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

  Data Pre-processing

  For the test data daily dispatch count is to be predicted. So first build the catboost regressor model to predict the daily dispatch count and then by using the daily dispatch count build the final catboost regressor model for weekly dispatch count prediction.

Daily Dispatch Model Feature Importance

Weely Dispatch Model Feature Importance

Daily Dispatch Count Prediction Plot

Weekly Dispatch Count Prediction Plot