Forecasting Salary Ranges for IT Professional in Marketplace Employing The Support Vector Machine Technique

Abstract

The advancement of the digital sector in Indonesia has resulted in a heightened demand for IT professionals. Additionally, there is a requirement to analyze and outline salary levels according to job profiles to foster transparency and efficiency during recruitment. This research seeks to forecast salary categories for IT positions utilizing the Support Vector Machine (SVM) technique at prominent marketplace companies, including Gojek, Shopee, Tokopedia, Traveloka, Tiket.Com, and Bukalapak. The dataset utilized comprises 611 records and features attributes like company, work location, experience, skills, and salary. The preprocessing steps include label encoding, numerical normalization, and multi-hot encoding for the skills features. Salary categories are classified into three groups: low, medium, and high. The SVM model is trained using a Radial Basis Function (RBF) kernel and assessed with metrics such as accuracy, precision, recall, and f1-score. The evaluation outcomes indicate that the SVM model effectively categorizes salary levels with an accuracy of 82%. This model exhibits its strongest performance in the Medium salary category, achieving an f1-score of 0.93. This research demonstrates that SVM can serve as an efficient alternative for developing a prediction system for IT salary categories.