Machine learning for retention

Staff turnover averages 63 per cent across the Australian meat processing industry. Retaining experienced employees continues to be one of the greatest ongoing challenges within the sector, directly related to higher costs and lost time as well as indirectly to higher incident accident rates. 
 
This project explored the possibility of analysing human resources data to predict employee retention risk. The goal was to enable processors to intervene in a targeted way to raise retention rates. Six algorithm models were tested for their ability to determine which employees were likely to have left the business after 12 months. One was identified as the most successful and two additional processors were brought on in addition to the original processor stakeholder to supply test data.
 
Between 59 per cent and 87 per cent of those staff identified as likely to leave did in fact leave. Results indicate the machine learning model is a viable tool for reducing turnover in red meat processing plants.