Implementation of Machine Learning-Based Classification Model in Employee Recruitment Decision Prediction

Employees are vital assets for any organization, and accurate recruitment decision-making is crucial for the organization's long-term success. Incorrect decisions can lead to high costs due to re-hiring processes, onboarding, and decreased productivity. This study aims to develop a recruitment decision prediction model using data obtained from the Final Results of the 2024 CPNS Recruitment in the Ministry of Finance. The data includes attributes such as educational background, age, GPA, SKD Score, and SKB Score. To understand the relationships between variables, correlation analysis was conducted using a correlation matrix and heatmap visualization. Additionally, data exploration was performed using histograms to show the influence of attributes on recruitment decisions. This study employs five machine learning algorithms for prediction: Linear Support Vector Machine, Decision Tree (C5.0), Random Forest, k-Nearest Neighbor (k-NN), and Naïve Bayes Classifier. The results indicate that some attributes significantly influence recruitment decisions, and machine learning models can identify candidates who are more suitable for the available positions. Among the five models tested, Naïve Bayes proved to be the most effective, achieving an accuracy of 88% and an AUC of 0.97, demonstrating its strong performance in distinguishing positive and negative classes. The key factors contributing to the model's success include relevant feature selection, data quality, as well as appropriate preprocessing and validation techniques. This model is expected to enhance objectivity, efficiency, and accuracy in employee recruitment processes, thereby assisting organizations in making more precise and fair decisions.