Abstract Every company, regardless of location, industry, or size, struggles with the problem of employee turnover and attrition. Predicting employee turnover is one of the top priorities for the human resources departments of many businesses because it is such a significant challenge. Employee turnover costs organizations a lot of money. In this research, we implemented multiple machine-learning methods to create a model that predicts employee attrition. Among them, the CatBoost algorithm is incorporated to identify a suitable approach for predicting employee attrition tasks early. The primary purpose is to find a method to predict the number of employees leaving their jobs accurately. Following training, the model for predicting employee attrition is assessed using a real dataset provided by IBM Analytics. This dataset has 35 features and around 1500 samples and is used to evaluate the model. Using CatBoost, we got a high accuracy on the Kaggle dataset titled "IBM HR Analytics Employee Attrition & Performance." We recommend using a technique called "synthetic generation" to create more combined features based on arithmetic operations, which improves the accuracy and area under the curve (AUC) of the original CatBoost model. This will allow you to get the most out of the fundamental characteristics of the dataset. We achieved high accuracy of 95.84% and consumed less time of 2.15sec as related to relevant studies; this indicates that our method is effective.
Alan : Mühendislik
Dergi Türü : Uluslararası
Benzer Makaleler | Yazar | # |
---|
Makale | Yazar | # |
---|