Predicting agri-food prices with time-series and data-mining based methods

Abstract

Onion, potato and tomato (OPT) are some of the most essential commodities of every Indian household due to their affordability and need. But most of the times it has been seen that the prices fluctuate throughout the year and sometimes affording some of these items become difficult for a typical middle-class family in India. Especially, the case of onion, being a food item in most of the households, the market behaves very erratically and is a matter of concern. There is hardly any concrete evidence available why the price of onion fluctuates drastically leading to consumers' grief. In this paper, the price of onion has been modeled using Winters' and (S)ARIMA(X) (auto-regressive integrated moving average method with seasonality (S) and exogenous variable (X)) algorithms using Python. The paper has tried to predict the prices and cites some of the causes for such fluctuations. It is observed when supply/arrival has no major role in the price fluctuations; there the volatile wholesale price is statistically significant causing the higher price fluctuations. The question behind high wholesale price fluctuations, given constant/relatively constant supply, is a matter of further supply chain research.