OPTIMIZATION OF SUPPORT VECTOR REGRESSION PARAMETER BASED ON FIREFLY ALGORITHM AND GENETIC ALGORITHM FOR FORECASTING STOCK PRICES IN CONSTRUCTION AND BUILDING SECTOR Erlyne Nadhilah Widyaningrum, Irhamah, Heri Kuswanto
Institut Teknologi Sepuluh Nopember
Abstract
Stocks are strongly supported by the government and a law has been passed regarding the implementation of activities in the capital market sector. By including capital or stock investment, investors have claims on the assets and income of a company. Stock prices fluctuate and tend to be dynamic at any time, so stock price predictions are needed to maximize profits for investors and avoid losses due to the nature of stock prices. Stock price closing data is generally non-linear, so the SVR (Support Vector Regression) model is used which offers an optimal global solution that works by mapping data to a high-dimensional space and has good performance in solving time series problems and non-linear data. However, to get optimal SVR results, it is necessary to choose parameter values carefully or optimize them so that local optimum values are not obtained, so in this study genetic algorithm optimization and firefly algorithm optimization methods were used in selecting SVR parameters to obtain better forecast results. So, from these calculations, this study found that optimizing the SVR parameters using the genetic algorithm and the firefly algorithm resulted in better forecasting performance.
Keywords: Support Vector Regression, Optimization, Firefly Algorithm, Genetic Algorithm
