jolly-io / Modeling_Customer_Churn_Prediction

In this project, the objective was to build a classification model to predict a customer’s likelihood to churn for ZQ, a telecommunications company.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Modeling_Customer_Churn_Prediction

Executive Summary amd Insights:

In this project, the objective was to build a classification model to predict a customer’s likelihood to churn for ZQ, a telecommunications company. Applying the data mining process to the dataset that was provided, I implemented data exploration, preprocessing and dimension reduction. I used a 10-fold cross-validation technique to assess model performance and determine among two models (Logistic Regression and Classification and Regression Trees (CART) which was the most effective model to deploy. On the basis of model performance, I decided to deploy the Logistic Regression model for the churn classification problem. Upon deployment I recorded, reported and interpreted the findings from the model to elicit key recommendations for our client, ZQ. Based on our findings and insights, I outline our managerial recommendations for our client as follows:

  • Focus on addressing the needs and concerns of senior citizens to reduce their likelihood of churn.
  • Incentive customers to opt for longer-term contracts, as they significantly reduce churn.
  • Improve the quality and reliability of the fiber optic internet service to mitigate churn among customers using that service.
  • Strengthen online security, device protection, and tech support services to retain customers who value these features.
  • Enhance streaming TV and streaming movie services to improve customer retention.
  • Investigate reasons behind the higher churn among customers with multiple lines and develop strategies to enhance their satisfaction and loyalty.
  • Customers who use electronic checks for payment have a slightly higher likelihood of churn compared to those using other payment methods. Telco could consider promoting alternative payment methods to reduce churn among this group of customers and also evaluate paperless billing systems to ensure a positive customer experience and explore ways to address concerns associated with it.
  • Customers who prefer mailed checks for payment have slightly lower odds of churning compared to customers using other payment methods. While not statistically significant, ZQ can continue offering mailed checks as an option, but it should also focus on promoting more convenient and efficient payment methods.
  • Monthly charges have a weak negative association with churn likelihood. As monthly charges increase, the odds of churn slightly decrease. ZQ can leverage this finding by ensuring that customers perceive the value in relation to their monthly charges.
  • Total charges have a significant negative association with churn likelihood. As total charges increase, the odds of churn decrease significantly. ZQ should emphasize the benefits and value customers receive as their total charges increase to strengthen customer loyalty and retention.

Importantly, I would recommend introducing end-to-end metrics to measure the effectiveness of any customer retention initiative ZQ proceeds to implement, following this report. This is to ensure agile feedback and iterative improvements where necessary. In conclusion, by considering these interpretations and implementing appropriate strategies and initiatives, ZQ can optimize payment methods, pricing, emphasize customer value propositions and improve other specified areas of its business to reduce churn and improve customer retention.

About

In this project, the objective was to build a classification model to predict a customer’s likelihood to churn for ZQ, a telecommunications company.


Languages

Language:R 100.0%