Optimasi Algoritma Prophet untuk Prediksi Harga Saham

Abstract

This research aims to improve the accuracy of daily stock price prediction using prophet algorithm with parameter optimisation method. Stock price prediction is an important part of investment decision making, where prediction accuracy can help reduce the risk of loss and increase profit opportunities. Stock prices are often influenced by various internal and external factors, such as global economic conditions, government policies, and market sentiment, making their movements highly volatile and difficult to predict precisely. For this reason, there is a need for machine learning algorithms that are efficient in modelling complex seasonal patterns and trends. prophet is a time series-based model that is effective in handling such patterns, but its accuracy can still be improved through hyperparameter adjustment. In this study, daily stock data of PT Telekomunikasi Indonesia Tbk from 13 July 2020 to 12 July 2024 obtained from Yahoo Finance was used, and parameter optimisation was carried out using the random search method to find the best parameter combination. The results show that the optimisation process is able to significantly improve prediction accuracy compared to the basic model without optimisation, with a clear decrease in predictive error from 10.28% to 2.94%. This research contributes to the development of a more accurate and efficient stock prediction model, and can be used as a reference in making decisions in finance and investment.