Applied aspects of time series models for predicting residential property prices in Bulgaria

Abstract

Accurate housing price forecasts play a critical role in balancing supply and demand in the residential real estate market, as well as in achieving the goals of various stakeholders – buyers, investors, construction contractors, public administration, real estate agencies, special investment purpose companies, etc. The present study aims to investigate the relationship between specific predictors and build a suitable model for forecasting housing prices in Bulgaria. In this regard, a study was conducted on transactions with residential real estate in the city of Sofia for the period from the first quarter of 2016 to the fourth quarter of 2021. The ARIMA model is used in the development to predict the values of the variables. Eight models are tested for the researched factors (24 in total). On this basis, the price per square meter of residential property was predicted, including estimated values from the ARIMA model for the parameters involved in the regression equation. The result showed that there is a strong relationship between the analyzed predictors and the studied variable – price per square meter of housing. The tested models are adequate and the statistical requirements for forecasting the prices of residential properties in Bulgaria are complied.

Comparative analysis of different approaches to the Ukrainian residential property market evolution modelling and its forecast for the years 2019–2024

“…The ARIMAX model, which combines the strengths of ARIMA with exogenous variables, offers a robust framework for modelling and forecasting time series data, accounting for both temporal patterns and external influences. While ARIMA models capture intrinsic temporal dependencies, they may overlook the impact of external factors on the target variable (Iliychovski et al , 2022). The integration of exogenous macroeconomic factors into the model addresses this limitation.…”

“…Their study underscored the strong desire for homeownership among the Polish population and emphasized the role of interest rates and credit availability in driving housing purchases. Iliychovski et al (2022) conducted a significant study focused on investigating the relationship between specific predictors and building a suitable model for forecasting housing prices in Bulgaria. The models were assessed for predicting the price per square meter of residential property, incorporating estimated values from the ARIMA model for the parameters involved in the regression equation.…”

Machine Learning and Data Balancing Methods for Bankruptcy Prediction

“…Machine learning methods are used to solve various economic problems. For example, research (Geldiev et al 2018) focuses on applying machine learning to build an accurate predictive model; debt management is assessed using support vector machines (Zakhariyev et al, 2020), consumer behavior of food retail chains is clustered using machine learning algorithms (Lyashenko et al, 2021); house prices in Bulgaria are projected using time series models (Iliychovski et al, 2022).…”