Assessment of Predictive Models for Coffee Production in the Philippines

Abstract

This is a research-in-progress of developing a coffee eco-market with online bidding for different coffee stakeholders in selected provinces in the Philippines. The objective of this paper is to compare three different forecasting models using a five-year coffee production data. The three models explore and assess exponential smoothing, moving average, and regression. Different components such as seasonal, trend and irregular components are present in the data. Thus, the original data is adjusted by removing the seasonal component, trend component, and irregular component. For the computation of the forecasted values, the MS Excel data analysis tool is used. The standards used to measure the accuracy of each three model for comparison are the MAE, the MSE, and the MAPE. Among the three model, the moving average model rank first with a 9% error accuracy percentage, the next is the exponential smoothing with 12% error accuracy percentage, then the last is the regression with 14% error accuracy percentage.