Pengembangan Algoritma Genetika dan Self Organizing Maps (SOM) untuk Prediksi Saham pada Emiten yang Terdaftar di PT. Bursa Efek Indonesia

Abstract

The development of methods or tools to determine stock price movements is currently being developed to minimize risks for stock traders and investors in making investment decisions on issuers listed on the Indonesian stock exchange. One method that is developing rapidly is the machine learning method. In this research, we develop a machine learning method with function optimization combined with a Genetic Algorithm which is able to increase the accuracy and prediction of an algorithm model. In the process of designing a machine learning algorithm with an artificial neural network function, namely the SOM (Self Organization Map) algorithm with the addition of a Reference Vector which is optimized with a Genetic Algorithm to predict the issuer's share price. The process of developing the SOM (Self Organization Map) algorithm with the addition of a Reference Vector optimized with the Genetic Algorithm will be built on a website-based program that can visualize starting from share price movements per company, be able to visualize the SOM algorithm, and be able to predict share prices for a maximum of 100 days in the future. . The results were obtained from 11 company sectors on the Indonesian Stock Exchange, namely, Health, Raw Materials, Finance, Transportation & Logistics, Technology, Non Cyclical Consumers, Industry, Energy, Consumer Cyclicals, Infrastructure, Property & Real Estate. From the results of research on datasets that have been trained and tested as well as evaluation results using this method, the energy sector is included in the sector that has the highest profits compared to other sectors. The energy sector saw profits increase by 11%. Meanwhile, the sector that experienced losses when investing was the Consumer Cyclical sector, where this sector experienced a decline in profits of negative -40%, with the lowest error value found in the property and real estate sector where the average Root Mean Squared Error (RMSE) value was 28 .8, Mean Absolute Error (MAE) 27.8, and Mean Absolute Percentage Error (MAPE) 17.4.