Deep Learning Approach for Prediction Employee Attrition for HR Analytics

Employee attrition has been a major challenge for organizations as it influences the productivity and the continuity of operations within any organization. The objective of this paper is to identify some of the crucial factors that lead to attrition through the analysis of employee data by providing an effective model that can be used by HR for decision-making. This paper focuses on an analysis of employee attrition where traditional ML techniques are used in tandem with advanced deep learning models for predicting and understanding the factors leading to employee turnover. In this paper, traditional ML models such as Logistic Regression and Random Forest Classifier, while being used in combination with a voting classifier ensemble and an integrated hybrid deep learning model based on CNNs and LSTM networks. Scaling of numerical features and one-hot encoding for the categorical variables will be implemented to ensure that the data is appropriately structured for each model to apply 'HR-Employee-Attrition.csv'. The Voting Classifier leverages predictions from multiple machine learning models to increase the possibility of providing an accurate answer. The deep learning model, using CNN layers but further enhanced with LSTM layers, can identify structural patterns as well as further temporal patterns within employee data. Using the models, a performance metric of accuracy score was 0.87, ROC-AUC was 0.54, along with confusion matrices to compare the deep learning model with other models for a better performance in terms of predictive power.