An Algorithm for Predicting Coffee Prices Using ARIMA Model

Abstract

Purpose: In this study, an algorithm for predicting coffee prices was developed incorporating the ARIMA model. The Algorithm simplifies the process of finding the optimal parameter values (p,d,q) for the ARIMA model. Methodology: Secondary data from the Uganda Coffee Development Authority monthly reports was used. The data involved monthly coffee prices of Arabica and Robusta coffee for the years 2014-2021. CRISP-DM methodology and Python programming language were used. Arabica and Robusta coffee prices for the years 2019-2022 were predicted. Findings: The study showed seasonality in the prices. Unique Contribution to Theory, Practice and Policy: The study recommends that coffee farmers, traders, cooperatives, and the Uganda Coffee Development Authority use the forecasting tool, link it to market information platforms for easy access to regular price updates, and enhance it to track seasonal price changes.