Genetic Feature Selection Applied to KOSPI and Cryptocurrency Price Prediction

Abstract

Feature selection reduces the dimension of input variables by eliminating irrelevant features. We propose feature selection techniques based on a genetic algorithm, which is a metaheuristic inspired by a natural selection process. We compare two types of feature selection for predicting a stock market index and cryptocurrency price. The first method is a newly devised genetic filter involving a fitness function designed to increase the relevance between the target and the selected features and decrease the redundancy between the selected features. The second method is a genetic wrapper, whereby we can find the better feature subsets related to KOPSI by exploring the solution space more thoroughly. Both genetic feature selection methods improved the predictive performance of various regression functions. Our best model was applied to predict the KOSPI, cryptocurrency price, and their respective trends after COVID-19.

Genetic Algorithm for Feature Selection Applied to Financial Time Series Monotonicity Prediction: Experimental Cases in Cryptocurrencies and Brazilian Assets

“…Methods such as meta-heuristic searches are widely adopted in the feature selection process in order to obtain satisfactory results [ 9 , 11 , 18 ]. These methods, instead of finding an optimal solution to the problem, aim to obtain a reasonable solution in a shorter period of time.…”

“…Several works used GAs to perform feature selection on financial time series data [ 9 , 11 , 18 ]. Cho et al [ 9 ], for example, proposed the use of a GA for feature selection to predict the Korean KOSPI index and the Bitcoin cryptocurrency. Feature selection was performed using a filter and wrapper method.…”

“…The various scientific studies that have been carried out on predicting the time series of financial assets are evident [ 2 , 9 , 10 , 11 , 12 , 13 , 14 ]. According to Cho et al [ 9 ], there are two most commonly used ways to predict the future value of these series.…”

“…The various scientific studies that have been carried out on predicting the time series of financial assets are evident [ 2 , 9 , 10 , 11 , 12 , 13 , 14 ]. According to Cho et al [ 9 ], there are two most commonly used ways to predict the future value of these series. The first focuses on improving feature selection methods, that is, it aims to find the best set of variables capable of explaining the movement of the time series.…”

Migration-Based Moth-Flame Optimization Algorithm

“…During past decades, optimization techniques have been developed widely to solve complex problems that emerged in different fields of science, such as engineering [1][2][3][4][5][6][7][8][9], clustering [10][11][12][13][14][15][16][17][18], feature selection [19][20][21][22][23][24][25][26][27][28], and task scheduling [29][30][31][32]. Such optimization problems mainly involve characteristics such as linear/non-linear constraints, nondifferentiable functions, and a substantial number of decision variables.…”