CC BY
13INNOVATION APPROACH TO ONLINE FORECASTING THE DYNAMICS OF HOUSE PRICES IN
UKRAINE
Chernenko V.,
Kremenchuk Mykhailo Ostrohradskyi National University, Association Professor of Department of Computer Science and Highest Mathematics, Ph.D. (Physics and Mathematics)
Slon Ya.
Kremenchuk Mykhailo Ostrohradskyi National University, Master's Degree of Department of Computer Science and
Highest Mathematics
Abstract
Forecasting prices, supply and demand for housing is the final and most important stage in the research of the housing market and is the construction of a scientifically grounded scenario of possible development of the housing market. This paper aims to predict prices in the housing market on the basis of the perspective method of prediction of dynamics series - the method of mean square collocation. This development allows real estate agencies, construction companies and other enterprises whose activities are related to investing in real estate to automate the evaluation and forecasting of real estate prices.
Keywords: information web system, real estate market, housing market, house prices forecasting prices, mathematical model of a collocation.
INTRODUCTION
In the modern world, a lot of attention is paid to the modeling of various economic and technological processes. Notably, there is a sufficiently large number of statistical data covering almost all spheres of human activity. This too applies to the housing market. However, looking only at figures from a table or a graph, it is difficult to determine the exact fluctuation of the house prices in a given period.
Prediction of price, supply and demand is an essential element of the economic and mathematical modeling and is the construction of a scientifically grounded scenario of the possible development of the housing market: the forecast of future sales, the identification of possible volumes of construction, and such like. The forecast of the housing market gives an idea of what will happen to the housing market in the future. This can help a user with their investment decisions.
The state of the current housing market is relevant and should be researched. The cost of housing, which is reflected in average prices in the primary and secondary housing markets, is a key factor for forecasting prices in the housing market. Using average prices in the regions, prices are calculated in the housing market at the state level as a whole. These indicators are taken into account when predicting the market price of housing, which is used in:
- social policy planning (calculation of various payments, budget funds for the construction of residential premises, etc.);
- implementation of the comparison of regions (regional policy to ameliorate differences in the economic development of regions);
- the development by government of strategic actions for the development and improvement of the housing sphere;
- calculating tax revenues and budgets of different tiers of government, etc.
The above aspects reflect the relevance of the chosen research direction.
Many studies, by both domestic and foreign authors, are related precisely to the allocation of factors that determine prices in the residential real estate mar-ket.For example in the paper [1], the authors created a predictive mathematical model of the dynamics of changing the prices of housing using neural network technologies based on the computing systems of nonlinear dynamics of time series, financial ranks of the price level of housing and macroeconomic and financial indicators of the country. In the paper [2] a regional analysis of the housing market in Ukraine was carried out: on assessment of the relevant factors, regions of Ukraine were clustered according to their house prices.
In the study [3], the authors examine the role of the real estate market in the Greek economy and also provide macroeconomic factors such as mortgages in the retail sector of the real estate market and show their correlation with the house price index. In the analysis of the UK's housing market, the authors of the paper [4] used the co-integration approach and the error correction model, which shows that the growth of house prices is most influenced by the interest rate. The results of research [5] have shown that new build prices in Serbia are increasing with population growth and real wages.
Researchers who deal with the problems of forecasting prices in the real estate market of Ukraine, build static predictions and use rather cumbersome techniques that cannot be repeated by real estate profession-als.For example, the article [6] deals with the existence of advanced and delayed indicators of the state of the housing market. The research demonstrates that the fluctuation of housing prices is nearly always synchronized to the cycle of business activity, the former usually ahead of the latter with a small lag. In the work [7], when forecasting prices for the primary residential housing market in Kyiv, it is proposed to divide the existing range of the dynamics into separate periods, demonstrating a clear tendency to price increase or de-
crease, which in turn requires a large number of additional calculations. There is, therefore, a demonstrable need for more convenient universal models of forecasting.
In the paper [8], the authors suggest that a dynamic factor model be used to predict housing price inflation for five metropolitan areas in South Africa. The article [9] deliberates the effectiveness of the dynamic factor model as compared to the classical autoregressive models.
Note, that this method makes it possible to build accurate static predictions, but, as the processes in the real estate market are quite dynamic, there is a need to dynamically forecast housing prices in the coming months. Another problem is the inappropriateness of the methods developed in countries with a well-developed market economy to predict in countries with economies in transition.
Therefore, in this paper it is proposed to model the dynamics of house prices in Ukraine with the aim of online forecasting with the help of an economic-mathematical model based on the method of mean square collocation [10]. Using the model of mean square collocation, we can observe and predict variables according to their physical, economic or mathematical nature [11], which means that changes in house prices in the coming months can also be predicted. The collocation model of forecasting preserves the main advantages of the classical regression models. It can be used for constructing the optimal forecast of with reference to homogeneous data and can equally be applied for evaluating any peculiar characteristics on the basis of heterogeneous information.
Thus, the research goal of the work is to develop an informational web system using modern programming technologies for storing and online forecasting
In order to test the implementation of the method of mean square collocation and the efficiency of the information web system as a whole a set of statistical data is selected. Information about the initial set of statistical data: region - Kyiv; parameter - the average price per square meter, currency - dollar, approximation - sliding average (5 points), forecast period - 6 months.
statistical data on the dynamics of house prices in Ukraine.
MATERIALS AND METHODS
Presently, there are several web-portals, for example, [12, 13], which enable the viewing of statistics on real estate sales in a separate regions of Ukraine in the form of charts. But for further analysis of statistical data, for example, for prediction, it is necessary to store data in a format that is convenient for future calculations.
To accomplish the task, an informational web system «Dynamics and forecasting of house prices in Ukraine» was developed using modern programming technologies MVC, the Laravel framework and object-oriented programming in PHP. The developed web system is an informational structure available on the Internet under the domain name hpd-forecasting. info. The paper [14] describes the stages of designing an informational web application for storing statistics on the dynamics of house prices in Ukraine, in a format suitable for further analysis.
RESULTSAND DISCUSSION
The statistical data, for the developed information web system, is obtained from the web page [15] in the json format and stored in the database for further use during online forecasting. The stored data then allows a user to see a time series of monthly average house prices (per square meter) in a selected region of Ukraine.
The statistical data for 2006-2019 in Kyiv are taken for research (Fig. 1).
The results obtained following the simulation as described above by the developed information web system provide the online forecasting data in housing prices in the form of a table or a graph. Fig. 2 shows a screenshot of the website page with the simulation results in the form of a table, in Fig. 3 - as a graph.
111111111111111111111111111111111111
111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111
# Ç? Ç? ^ ^ \> \> O O O O ^ ❖ ❖
^ ^ ^ ^ ^ ^ # ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^
Fig. 1. Statistics on the dynamics of house prices in Kyiv from June 2006 to February 2019
3500
3000
2500
2000 _
1500
1000
500
Average price in Kiev
August 2019
July 2019 1141,31
June 2019 1113,66
May 2019 1083,70
April 2019 1054,08
March 2019 1027,61
February 2019 1044,00
January 2019 1037,00
December 2018 1039.0D
November 2018
Fig. 2. Screenshot of the developed web site page with the results offorecasting house prices in Kyiv in the form
of a table
Fig. 3. Screenshot of the developed web site page with the results offorecasting house prices in Kyiv in the form
of a graph
In Fig. 2 bold text indicates the months for which the projected values of average house prices per square meter in the chosen region are being obtained (from March 2019 to August 2019). The same information is provided in Fig. 3 in the form of a graph (the predicted values are constructed in the right half of the graph).
To test the adequacy of the constructed collocation model for the purposes of predicting average prices per square meter in the Ukrainian housing market it is necessary to compare the projected values for the twelve months - from March 2018 to February 2019, - with the actual values for this period.
In Table 1 demonstrates both the estimated average selling price of 1 sq. m. based on the model of collocation, and the actual average sales data of 1 sq. m. of
the residential real estate in Kyiv for the specified period.
The above exercise of adequacy verification demonstrates the proposed model's high accuracy, with
average error of approximation A = 2.8% being
relatively small (up to 10 %).
The projected values for large regions, such as Kiev, will be significantly improved if they form a sample by district of Kyiv. For example, for the Darnytsky district of Kyiv, the average approximation error is 1.6%.
Consequently, the high accuracy of the model of RMS can allow it to predict housing market prices in the short-term (up to one year) and in the mid-term (from one to three years).
Table 1
Verification of the Adequacy of the Collocation Model
Period Factual price, $ per sq. m Estimated average selling price, $ per sq. m Error of approximation, %
March 2018 1206.8 1219.2 1.0
April 2018 1164.7 1220.4 4.8
May 2018 1111.5 1190.4 7.1
June 2018 1087.4 1157.8 6.5
July 2018 1062.5 1125.4 5.9
August 2018 1062.7 1096 3.1
September 2018 1075.2 1071.4 0.4
October 2018 1061 1042.2 1.8
November 2018 1050.5 1041.5 0.9
December 208 1039 1050.4 1.1
January 2019 1037 1041.5 0.4
February 2019 1041.3 1050.4 0.9
CONCLUSIONS
The implemented informational web system provides access to the statistical information on fluctuations in house prices in any region of Ukraine, and facilitates its perception and analysis with the help of dynamic graphs. The results of the online forecast provide a basic idea as to how prices in the housing market will behave in the coming months. The uniqueness of this development is to enable users to store statistics in the json format and to dynamically receive results of online forecasting of housing prices in any region of Ukraine.
The results of this study may be useful to ordinary home buyers or investors who own funds in real estate, saving capital from inflation or for profit when renting a home.
The projections constructed in this paper are point-based and the authors believe that in the future interval forecast calculations should be added to allow for interval searches of the house prices in the regions of Ukraine. It is also planned to improve the forecasting process by taking into account the detailed factors associated with the formation of prices in the Ukrainian housing market.
REFERENCES:
1. Воронш В.О., Лянце Е.В., Мамчин М.М., 2010. Дослвдження ринку нерухомостг Проблеми, тенденци, прогнозування. Вюник Нацюнального ушверситету <^bBiBCbKa полггехшка». Серiя: Лопстика 690: 540-552.
2. Мезенцева Н.1. Дмитрieвa К.£., Кучевсь-кий А.А., Стеблецька Ю.Ю., 2012. Ринок житла Украни: суспiльно-геогрaфiчне дослщження. Еко-TOMi4ra та сощальна геогрaфiя: науковий журнал 64: 137-146.
3. Panagiotios T., Printzis P., 2016. On the mac-roeconomic determinants of the housing market in Greece: a VEMC approach. International Economics and Economic Policy 13(3): 387-409.
4. Xu L., Tang B., 2014. On the determinants of UK house price. International Journal of Economics Research 5(2): 57-64.
5. Nicolic I., 2015. Price determinants of newly dwellings in Serbia. Industrija 43(2): 105-116.
6. Шаповалова В.О., 2015. Прогнозування цш на нерухомють з урахуванням тенденцш дшових циктв. Економша та держава 9: 37-41.
7. Шапошшкова 1.О., 2018. Аналiз часових рядiв первинного ринку житлово1 нерухомостi м. Киева. Економiчний вiсник унiверситету 36/1: 139147.
8. Yang J., Liu H., Leatham D.J., 2013. The multi-market analysis of a housing price transmission model. Applied Economics 45: 3810-3819.
9. Das S., Gupta R., Kabundi A., 2010. Forecasting regional house price inflation: a comration between dynamic factor models and vector autoregressiv models. Journal of Forecasting 30(2): 288-302.
10. Бабешко Л.О., 2000. Коллокационная модель прогнозирования количественных характеристик основных финансовых инструментов фондового рынка. Вестник финансовой академии 2(14): 79-87.
11. Бабешко Л.О., Бывшев В.А. Прогнозирование финансово-экономических показателей по разнородным данным. Москва, 2017. 356 с.
12. Цены на недвижимость в Одессе. Одесса, 2019. URL: http://www.atlanta.ua/analytics/prices.
13. Портал недвижимости Харькова. Харьков, 2019. URL: http://sdelka.kharkov.ua/analitic.
14. Слон Я.В., Черненко В.П. Проектування шформацшного веб-додатку для збереження статистичних даних про динам^ цш на нерухомють. Матерiали II Мiжнародноï науково-техшчно1 конференцп «Комп'ютерш та шфор-мацшш системи i технологи» (18-19 квггня 2018, м. Харшв). Харшв, 2018. 33-34 с.
15. Портал недвижимости. Динамика и статистика цен на недвижимость: анализ рынка недвижимости и обзор рынка недвижимости в графиках. Киев, 2019. URL: http://domik.ua/nedvizhimost/dina-mika-cen.html.
