Skip to content

The objective of this project to is to predict customer churn, loss opportunity and provide recommendations to the business team so the company can implement a customer persona in retention strategy and can monitoring throught dashboard interactive.

Notifications You must be signed in to change notification settings

archie-cm/Churn-Analysis-Ecommerce-Customer

Repository files navigation

Full Stack Data - Churn Analysis Ecommerce Customer

Problem Statement : The E-commerce customer churn rate is up to 80% compared with traditional business customer management (Wu & Meng, 2016). From dataset customer churn is 17%.

Goals : The purpose of this project is to predict customer churn, loss opportunity and provide recommendations to the business team so the company can implement a persona customer retention strategy and can monitoring throught dashboard interactive.

Result : The results show that tenure and complaint has the greatest impact on a churn rate. With applying an actionable recommendation from insights, company can avoid loss oppurtunity up to $900,000 and revenue lift up to $ 150,000.

Tools: Python, JupyterLab, Git, PowerBI

Libraries: Pandas, Numpy, Feature-engine, Scikit-learn, Imbalanced-learn, kaplan-meier survival curve, lifelines.CoPHfilter, lifelines.predict_survival_function, K-means, gaussian, rfm-segmentation

Dataset: Ecommerce Customer Churn Analysis and Prediction [source]

Summary of the analysis

  • This dataset has 5630 observations and 20 variables with 15 numerical variables, 5 categorical variables and 2 target variable.
  • From the data visualization, it is obtained that the churn ratio has a correlation with tenure, complaints, cashback Amount, & preferred order cat.
  • The results of predicting churn are strongly influenced by the level of Tenure, Complaint, Number of Addresses, and cashback Amount.
  • The results of the Survival Analysis, the customer has the greatest survival chance in No Complain, Marital Status Married, Payment Mode Credit Card, Order Category Grocery. image
  • RFM Segmentation results show priority customer treatment in the Loyal, New, Promising, and Lost Potential segments. image
  • Total Expected Loss of $ 910,687
  • Estimated Revenue Uplift
    • Order category Grocery $42,448
    • Payment Credit Card $ 91,785
    • Payment Debit Card $ 78,543

image

image

What I have learned

  • Framing the business problem.
  • Create a machine learning model and extract insight that generates churn & retention from it to make an actionable recommendation for the business team.
  • Create a survival analysis, predict customer who will churn in Future and extract insight that generates churn & retention from it to make an actionable recommendation for the business team.
  • Create a customer segmentation with RFM Segmentation, KMeans and Gaussians that can generates strategy-strategy personal customer.
  • Create a dashboard interactive can monitoring business metrics, operational, and sales.
  • Make a business simulation from insights that calculate loss oppurtunity and revenue uplift.

File Dictionaries

About

The objective of this project to is to predict customer churn, loss opportunity and provide recommendations to the business team so the company can implement a customer persona in retention strategy and can monitoring throught dashboard interactive.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published