ARIMA-Based Time Series Modeling for Forecasting Egg Market Prices in Hyderabad

Resumen

Due to changing eating habits and increasing cost of pulses, the demand for protein-rich food
has increased, especially poultry products. Considering the rapid population and egg consumption of poultry
products, the country should increase its production. It is important to predict the future egg price with
the resources available today. This study aims to predict the price of eggs in Hyderabad by analyzing egg
price data from January 2019 to January 2025. The study focuses primarily on the time series model ARIMA
and also provides a performance analysis of the model. This work addresses a notable research deficiency
concerning the inadequate utilization of advanced error metrics and comprehensive model validation in prior
Indian studies on egg price forecasting. Previous research frequently neglects to account for variables such
as swift demographic transitions, changing food tastes, and market instability, all of which can significantly
impact the precision of predictions. Furthermore, there is a significant deficiency of long-term, city-specific
forecasting models designed to address the distinct difficulties and market conditions of the post-pandemic
era. This study seeks to address these deficiencies by integrating these factors to enhance the accuracy and
applicability of egg price forecasts in India. The ARIMA model is unique because it can use autoregression,
differencing, and moving averages in a flexible way. This makes it operate well with both stable (stationary)
and changing (non-stationary) time series data, which makes it more accurate at predicting future values in
a variety of fields. The accuracy of forecasting models was assessed by analyzing different errors. To test
the reliability of the model, (MAPE) mean absolute percentage error, (MSE) mean square error, R-squared,
(RMSE) Root mean square error and (BIC) Bayesian Information Criterion were used to test the reliability
of the model. The ARIMA (2,1,1) model was most appropriate for this dataset.