A lasso regression-based forecasting model for daily gasoline consumption: Türkiye Case

Abstract

Gasoline is one of the most sought-after resources in the world, where the need for energy is indispensable and continuously increasing for human life today. A shortage of gasoline may negatively affect the economies of countries. Therefore, analyzes and estimates about gasoline consumption are critical. Better forecast performance on gasoline consumption can serve the policymakers, managers, researchers, and other gasoline sector stakeholders. Parallel to the world economy, gasoline consumption in Turkey is among the top among the most consumed energy source. Therefore, it is aimed at forecasting the amount of daily gasoline consumption in Turkey in this study. For this purpose, a lasso regression-based forecasting methodology is proposed. The forecasting approach used for daily gasoline consumption consisting of 3 main stages: i) cleaning the data ii) extracting and selecting features iii) forecasting the future of daily gasoline consumption time series via the proposed models. Besides, Ridge Regression is used to compare the performance of the proposed model.