Track: Supply Chain Management

Abstract

Okra (Abelmoschus Esculentus) has a perishable nature and its variable prices as well as arrivals depend on the supply and demand in the market. Approximately 73% of worldwide Okra is cultivated in India. This research incorporates seasonal variation in the forecasting of arrivals and three categories of prices, which were deseasonalized to achieve a regression line. Initial values of level, trend and seasonality were exploited from the regression line and were used to perform static time series forecasting analysis. The same initial values were then updated for every new data entry for the adaptive time series forecasting model (Winters Model). The errors and fit of model were seen on the basis of MSE, MAD, MAPE and R2 values. A database for the arrivals as well as the maximum, minimum and modal prices of Okra was gathered for 12 years (Jan’10 to Dec’21). The static model gave a MAPE of about 19.33% for the prices and almost 207% for arrivals of Okra. On the other hand, the adaptive winter’s model, which incorporated the previous as well as the current data, streamlining itself after every new entry, gave an average error of 13.9% for the prices and about 18.9% for arrivals. This adaptive model predicted the modal price of Jan 2022 and Feb 2022 to be Rs. 5800/quintal and Rs. 5600/quintal, and the actual values observed are Rs. 5700/quintal and Rs. 5200/quintal.