Employee Turnover Prediction System

Employee turnover has been recognized as a main problem for industries due to its contrary influence on work place throughput and long term development tactics. To resolve this issue, officialdoms use machine learning techniques to guess employee turnover. Precise predictions qualify administrations to take action for holding of employees. Nonetheless, the data for this type of issue comes from HR Information Systems. This leads to the supremacy of noise in the data that extracts predictive models disposed to over-fitting and hence imprecise. This is the main problem that is the motivation of this study, and one that has not been handled earlier. The innovative contribution of this study is to reconnoiter the application of Gradient Boosting technique which is more robust because of its regularization formulation. The global retailer data is used to compare Gradient Boosting against three traditionally used supervised classifiers like Logistic Regression, Support Vector Machine, Random Forest and reveal its suggestively higher accuracy for predicting employee turnover. In this study I have also implemented Artificial Neural Network to study how to the neural network helps in classification of different classes. This paper is aimed at developing models which can predict employee turnover and it can help the organization to take necessary steps to retain these employees.