Skip to content

Latest commit

 

History

History
31 lines (18 loc) · 1.33 KB

Readme.Md

File metadata and controls

31 lines (18 loc) · 1.33 KB

Expresso-Churn-Prediction

Competition Link

DSN Pre-Bootcamp Hackathon: Expresso Churn Prediction Challenge by Data Science Nigeria This solution will help Expresso to better serve their customers by understanding which customers are at risk of leaving.

My Solutions was doing Exploratory Data Analysis, Created New columns from the TOP_PACK columns in the dataset.

I used three diffrent approach for my solutions Model using Kmeans to create cluster, Log Transfroming the Skewed Columns and No preprocessing with each given me a higher logloss than my baseline model of (Public LB: 0.2519)

Solution WorkThrough

  1. Exploratory Data Analysis
  2. Using Sklearn Label Encoder for Encoding Categorical Features
  3. Kmeans Clustering
  4. Ensembling

Improvement

  1. Feature Engineering
  2. Selecting the Besy K for the Kmeans Clustering
  3. Using Pandas Get Dummy Method for Encoding the REGION Features

My Score was a blending of the 3 Model (Log Transform Model, Model, Cluster Model)

Public LB: 0.25137 39/358

Private LB: 0.24713 27/358

If you have any questions or suggestions do not hesitate to contact me on linkedin