Predictive Modeling of Housing Construction in Russia: Insights From Arima and Local Regression Analysis

The purpose of this work is to identify the fundamental factors influencing the dynamics of residential real estate construction in order to further predict the population's housing supply.The problem that analysts face is the relatively short time series with annual data characterizing this market.Earlier methodological approaches to forecasting short-term series suggested the use of exponential moving averages, Holt-Winters modeling, ARIMA, LSTM neural networks, LOESS and others, mainly in Europe and the USA.In this study, the authors present the results of forecasting the dynamics of residential real estate construction in Russia using the ARIMA model, as well as a variety of series decomposition (STL) methods and taking into account all the rules of the Holt-Winters model.