A1414
Title: Voting-based ex ante method for selecting strategy of the price characteristics prediction on real estate market
Authors: Alicja Wolny-Dominiak - University of Economics in Katowice (Poland) [presenting]
Tomasz Zadlo - University of Economics in Katowice (Poland)
Adam Chwila - University of Economics in Katowice (Poland)
Monika Hadas-Dyduch - University of Economics in Katowice (Poland)
Tomasz Stachurski - University of Economics in Katowice (Poland)
Malgorzata Krzciuk - University of Economics in Katowice (Poland)
Abstract: The topic of selecting an optimal prediction strategy is addressed when utilizing parametric or nonparametric regression models. The term "prediction strategy" is understood as the pair: the assumed model and the prediction algorithm. It emphasizes the significance of ex-ante prediction accuracy, ensemble methodologies, and forecasting not only the values of the dependent variable but also a function of these values, such as the total or median. The research proposes a methodology for selecting a strategy to predict the vector of functions of the dependent variable using various ex-ante accuracy measures. The final decision is determined through a voting mechanism, wherein the candidates are prediction strategies and the voters are diverse prediction models with their respective prediction errors. As the method is based on a Monte Carlo simulation, it facilitates the consideration of novel scenarios not previously observed. First, a comprehensive theoretical description of the proposed method is provided, while subsequently, its practical application in predicting characteristics of prices on the real estate market is presented. The empirical example utilizes data from the USA market. All computational analyses are conducted using the R programming environment.
