YULIIA V. DEHTIAROVA, YURI YEVDOKIMOV
Data Mining Methods and Models for Social and Economic Processes Forecasting
(original language – English)
https://doi.org/10.21272/mer.2018.80.03
Developing social and economic systems and ensuring efficiency of social and economic processes is one of the major tasks for the government of any country. Forecasting models used for analyzing large data sets allow more efficient enterprise management. Big Data is a key resource that provides a competitive advantage to many businesses. The widespread use of Data Mining in retail, marketing, finance, healthcare, industrial production, and other areas suggests that gathered and processed information not only provides useful business information, but also allows more accurate evaluation and development of detailed business plans and development strategies. The use of Data Mining methods and models in forecasting tasks of socio-economic processes provides more accurate predictive calculations. To select the best method for solving prediction problems it is necessary to clearly understand whether the macro series is stationary or non-stationary. We must clearly understand whether our macro series is a clear trend or seasonality. For prediction of stationary time series, the most popular models are autoregression and moving average models. ARIMA models cover a sufficiently wide range of time series, and small modifications of these models allow seasoning time series to be more accurately described. This article discusses the role of Data Mining in social and economic processes, as well as the potential of using Big Data in a business environment. This article demonstrates procedures for using Data Mining methods for practical implementation of Data Mining algorithms to forecast Ukraine’s GDP with Python codes, using the statsmodels package. This article analyzes the possibilities of using ARIMA model and uses a double exponential smoothing model for forecasting Ukraine’s GDP.
Key words: Data Mining model, forecasting, socio-economic process, macro series, stationarity, trend, seasonality.
Placed in №2, 2018.
Affiliations:
Yuliia V. Dehtiarova, Student of the Faculty of Information Technology Taras Shevchenko National University of Kyiv.
Yuri Yevdokimov, PhD, Professor of Department of Economics, University of New Brunswick, Fredericton, Canada
Article text: download.