September 5, 2020
Leading retirement services provider
- A leading retirement services wanted to predict the attrition of its company for defined business units and defined bands of its employees
- The company was facing high attrition across its mid-level ranks and risks losing its key trained talent pool to other players in the industry
- Key employee attrition was a pain point and expectation was to build predictive models to understand attrition at different bands based on the internal and external factors
Analytics Led Approach
- Data cleansing and missing value imputation
- Time series data set created for modeling (quarterly
- Research on external variables influencing attrition
- Some of factors evaluated are like performance ratings, salary attitudes, designation, state wage index, resident status, department/BU etc.
- Explore various classification statistical techniques for an appropriate prediction model. (Decision trees, logistic and Support Vector Machine)
exploratory analysis on the data, where key factors impacting attrition were identified.
- A model was developed and delivered to predict the attrition over the next few quarters.
- A sensitivity (what if analysis) model was delivered for the consumption of the HR team to predict / forecast the attrition rate based on different internal (performance rating, bonus payouts, etc.) and external environmental factors (unemployment rate, stock index etc.)
Critical Success Factors
- Close involvement of stake holders during model development and validation phase
- HR team uses the model to predict the attrition rate of its key employees based on the different internal and environmental factors