Digital Skill Demand Mapping in Indonesia Using TF-IDF and K-Means Clustering

This preprint presents a study on digital skill demand patterns in Indonesia based on online job vacancy data collected from Glints Indonesia. The study analyzes 552 digital job vacancies using text mining and unsupervised machine learning techniques. The methodology consists of data preprocessing, skill extraction, Term Frequency-Inverse Document Frequency (TF-IDF) feature representation, K-Means clustering, and cluster evaluation using Silhouette Score, Davies-Bouldin Index, and Calinski-Harabasz Index. The analysis identifies SQL, Java, Python, Excel, PostgreSQL, MySQL, JavaScript, React, Cloud, and Node.js as among the most frequently demanded digital skills in the dataset. The clustering evaluation indicates that K=16 achieves the highest mathematical performance, with a Silhouette Score of 0.4868. However, K=14 is selected as the final configuration based on occupational interpretability and industry relevance. The resulting clusters represent fourteen digital occupation categories, including Data Analyst, Data Science, Data Engineering, Artificial Intelligence Engineering, Machine Learning Engineering, Backend Development, Full Stack Development, Frontend Development, Mobile Application Development, Database Engineering, UI/UX Design, Software Engineering, Android Development, and Quality Assurance Testing. This work provides an empirical mapping of digital workforce requirements based on online job vacancy data and may support further research in digital skills analysis, workforce development, curriculum alignment, and labor market intelligence. This document is a preprint and has not undergone peer review or been published in a journal at the time of deposit.