Institutional Repository of Vilnius University
Prekybos, maitinimo ir paslaugų sektorių mėnesinių pajamų analizė ir prognozavimas /
Abstract
dc:descriptionThis bachelor’s thesis explores the possibility of using time series forecasting models to improve the preparation of short-term statistics. The topic is relevant to public sector institutions that aim to optimize the process of producing statistical indicators, reduce the need for surveys, and make better use of already published data. Until now, the application of time series models to the estimates published on the Official Statistics Portal (OPS) has not been thoroughly examined — this thesis not only addresses this question but also suggests specific models suitable for forecasting. The aim of the thesis is to assess the suitability of ARIMA (AutoRegressive Integrated Moving Average) and SARIMA (Seasonal ARIMA) models for forecasting the monthly revenue of Lithuanian economic sectors. At first, the advantages and limitations of the i.SAF data source are discussed to assess its suitability for statistical analysis. Then, the OPS revenue estimates, aggregated according to the NACE Rev. 2 classification, are analyzed. A detailed analysis is carried out for one selected sector using the Box–Jenkins methodology — checking stationarity, applying differencing, analyzing ACF and PACF plots, and selecting a model both manually and automatically. For the other sectors, models are selected automatically using the auto.arima() function. Model accuracy is evaluated using AIC, RMSE, and MAPE indicators by splitting the data into training and testing sets. The impact of the COVID-19 pandemic on forecast accuracy was also assessed. For each sector, forecasts were created with and without pandemic data to evaluate how sensitive the models are to unexpected economic fluctuations. The results showed that ARIMA and SARIMA models provided accurate short-term forecasts in most cases, but their sensitivity to the pandemic period varied by sector — in some cases, removing pandemic data improved accuracy, but not always. The results can be used by the State Data Agency as an additional tool for producing early estimates when administrative data from other sources are not yet available.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2025
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Plančiūnaitė, Austėja Ona,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD239028998&prefLang=en_US