A Comparative Analysis of Time Series Models for Onion Price Forecasting: Insights for Agricultural Economics
Vinay H T *
Department of Agricultural Statistics, Uttar Banga Krishi Viswavidyalaya, Cooch Behar-736165, West Bengal, India.
Pavithra V
Department of Agricultural Statistics, Uttar Banga Krishi Viswavidyalaya, Cooch Behar-736165, West Bengal, India.
Jagadeesh M S
Division of Agricultural Economics, ICAR-Indian Agricultural Research Institute, New Delhi-110012, India.
G. Avinash
ICAR-Indian Agricultural Statistics Research Institute, New Delhi-110012, India.
Harish Nayak G H
ICAR-Indian Agricultural Statistics Research Institute, New Delhi-110012, India.
*Author to whom correspondence should be addressed.
Abstract
Onion price forecasting plays a critical role in agricultural planning, market stability, and consumer welfare. By predicting onion prices, stakeholders can make informed decisions regarding planting, harvesting, trading, and consumption, mitigating risks and ensuring sustainable supply chains. The present study aims to forecast monthly wholesale onion prices in Bangalore market by using various statistical techniques like Exponential Smoothing, ARIMA, SARIMA, BATS and TBATS models. The time series price data from January 2010 to December 2023 was utilized for the study. Models were trained on 80% of the data and validated on the remaining 20%. The performance of each model was compared based on the two metrics like Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE). Results revealed that the TBATS model was performing better as compared to other six models with less RMSE of 281.30 and MAPE of 0.1544 Additionally, the residuals of the TBATS model exhibited a random distribution. Therefore, the onion prices for January 2024 to December 2024 were forecasted using the TBATS model.
Keywords: Forecasting, ARIMA, SARIMA, BATS, TBATS