Customer Churn Prediction
Customer Challenges:
- To predict the corporate customers who are going to churn in next 3 months
- What are the drivers for the churn and how to reduce the churn?
Result:
- 8% reduction in Customer churn for the 2 quarters
- 5 % increase in revenue by Cross sell and up-sell
- Increase in Customer satisfaction
CoreView Solution:
- Analysis of 100K B2B customers data for last 24 months
- Feature engineering – Calls data, Billing data, Care tickets data etc
- Python – Data processing and time-series forecasting
- Models based on Random Forest, Clustering and LSTM
- ARIMA time series forecasting
Other Considerations:
- No direct measure for effectiveness
- Parameters and their impact on results is not well defined
