Forecasting Gasoline Retail Prices Using Predictive Analytics
Keywords:
ARIMA Model, gasoline retail prices, forecasting, predictive analytics, time series analysisAbstract
Forecasting gasoline and diesel retail prices is crucial for consumers, businesses, and policymakers due to the significant economic impact of fuel costs. This study aimed to address the challenge of predicting these prices using predictive analytics. It sought to identify weekly price trends, develop accurate forecasting models for various types of gasoline and diesel, and create an application to integrate these models. Data from Kaggle, encompassing 1,362 instances with attributes (date, regular, midgrade, premium, and diesel) over 26 years (1995-2021), were analyzed. The findings indicated inconsistent prices across gasoline types and a recent sharp decline in retail prices. Using the ARIMA (1,0,0) model, the study developed a forecasting model with minimal differences between actual and forecasted prices. The five-week forecast revealed a steady price increase, with a low RMSE of 0.22 to 0.26, indicating excellent model accuracy. Additionally, an application with a dashboard for visualizing weekly gasoline and diesel price trends was developed. This study successfully provides a model for forecasting retail gasoline prices, helping consumers anticipate price increases.