Best-Fit: Best Fit Employee Recommendation

Abstract

In this fast-growing world, there is huge competition in the market for employees. It becomes a tough task from an HR perspective to keep the most talented resources in the company to benefit the productivity of the company because every employee is a valuable asset. Employees tend to shift their current jobs numerous times due to various reasons and therefore employee turnover becomes a serious issue in this challenging world. This paper focuses on resolving the problem of employee attrition using classification algorithms like random forest, logistic regression and SVM on the IBM attrition dataset. If a valuable employee leaves an organization or gets promoted it becomes a difficult and tedious task to replace the employee. Architecture has been proposed in this paper which uses Random forest, SVM, Decision tree classifiers and similarity techniques to find the closest employees suitable for the vacancy. This includes finding similarities in the skills, qualifications and experience. Pre-trained word vectors are used to generate GloVe embeddings for finding document similarity. A personality match between two employees is calculated by taking a big five personality test, followed by clustering and finding the euclidean distance between two answer vectors in the same cluster. Best-Fit will finally recommend best-fit employees on the basis of resume match, personality match and retention probability.