Title Nowcasting lithuanian gross domestic product (gdp) using machine learning methods /
Translation of Title Lietuvos bendrojo vidaus produkto (BVP) prognozavimas realiuoju laiku naudojant maĊĦininio mokymosi metodus.
Authors Kulikauskas, Adomas
Full Text Download
Pages 59
Keywords [eng] Gross domestic product (GDP), Nowcasting, Machine learning, Gradient Boosting Regressor, ARIMA, Maximal overlapping discrete wavelet transform (MODWT).
Abstract [eng] Gross domestic product (GDP) is one of the main key indicators of a state's economy. Nowadays GDP rates are available only with weeks of delays, using nowcasting methods, GDP predictions are accessible in the first month of the quarter. In this work, a Lithuanian GDP nowcasting system was developed with ten different machine learning models. Also, data provided by the state data agency with appropriate data delays were considered. Practical implementations of nowcasting were introduced with a pseudo-real-time framework, where different data vintages performed different time periods of predictions. The improvements in model performance were introduced using maximal overlapping discrete wavelet transform.
Dissertation Institution Vilniaus universitetas.
Type Master thesis
Language English
Publication date 2025