Aplikasi Prediksi Harga Saham Telkom Indonesia Berbasis Streamlit Menggunakan Machine Learning

Abstract

Fluctuations in the share price of PT Telkom Indonesia (TLKM) are a major obstacle for individual investors, who generally rely on non-numerical analysis. This study attempts to create a simple prediction solution by comparing the effectiveness of three machine learning algorithms, namely linear regression, support vector machine (SVM), and neural network (NN), then applying them to an interactive web application using the Streamlit framework. Historical TLKM stock data was prepared through a feature engineering process before performance testing. According to the comparison results, the Support Vector Machine (SVM) model proved to have the best predictive ability compared to Linear Regression and NN. The superiority of SVM can be seen from the lowest RMSE (Root Mean Squared Error) value and the highest R² (Coefficient of Determination) score, which shows that SVM is better at capturing non-linear patterns and the complexity of the TLKM market. The selected model was then integrated into the Streamlit application, which provides real-time prediction results and comparison visualizations. This research successfully bridges the gap between the accuracy of advanced machine learning models and the real needs of investors by providing an informative and easy-to-use decision-making tool.