<![CDATA[The use of skincare products in Indonesia has increased significantly along with the increasing public awareness of the importance of skincare, but this also raises indications of dependence behaviour that needs to be anticipated, especially in young age groups. This research aims to build a skincare dependency prediction system based on demographic, psychological and behavioural attributes collected through an online survey. In addition, a comparison of five classification algorithms-Decision Tree, Random Forest, Support Vector Machine, K-Nearest Neighbors, and Logistic Regression-was conducted to determine the best model that is most accurate and efficient in predicting the dependency tendency. The data obtained was processed through normalisation and categorical feature transformation with One-Hot Encoding, then evaluated using accuracy, precision, recall, and F1-score metrics. The results showed that the Decision Tree algorithm provided the best performance with accuracy reaching 87% and excellence in model interpretability. The model was then implemented in the form of an interactive web application based on Streamlit that allows users to make predictions independently and in real-time. The contribution of this research is the availability of a prediction system that supports education and wiser decision-making in the use of skincare, as well as opening up opportunities for the utilisation of machine learning technology for other issues.]]>
Copyrights © 2025
