Talent Flow Employee Analysis based Turnover Prediction on Survival Analysis

Abstract

When an employee-departures form a company, it can result in a significant loss. Therefore, we are predicting employee turnover with the help of human resource management. There are some previous researches are happened but they are mainly focused on centered turnover of an employee. They are ignored past events of company’s turnover behavior and also the statistical data for each job. For that we are using an algorithm called CoxRF an employee centered turnover prediction, in this we are combining the statistical results of survival analyzing with the help of assembly learning to reduce the problem called conservative supervised binary classification for an event centered perspective. To help with the structure of survival data from censored data, we coined the terms "event-person" and "time event". We interrelate CoxRF to a number of baseline approaches using an original dataset of china’s largest technological social network. The result shows that it is a good turnover interpreter. The following discoveries have been made: i) Employee turnover varies by industry, with the IT sector having a slightly higher rate than the government sector; ii) Gender plays major role if it is a women after marriage some are relieving from work and other factors also; iii) The person with good academic records can work more efficient than the other with low; iv) GDP plays an vital role in turnovers of an company and an employee in the current situation which have been neglected in previous studies; v) And the final one is for salary hike they are relieving is one of the problem for that we are losing a good employee.