Hackathon: Decision Trees to Improve Staff Retention
At the end of 2019 I participated in a rapid 5-hour hackathon with the objective of using data science to improve staff retention.
Due to the sensitivity of employee data, we used a dataset based on a combination of anonymised and fictional data, compiled by HR (sufficiently realistic to test our hypothesis).
The challenge was to analyse the data, draw any insight from it, and develop a model to identify staff who were at a high risk of leaving.
Following some data exploration, we developed a decision tree model that was able to accurately predict leave risks, as well as the factors that were most likely to contribute to staff leaving.
I produced a series of recommendations in a quick slide deck, which you can view below.
Whilst the data was fictional, it was fun to apply analytics to a self-contained and novel problem. I believe we demonstrated that data science could benefit HR and improve staff retention, bringing significant business value.
The HR team are now setting up a small project to take forward the recommendations, and to test the model on real-life data.
See my presentation here.
