Statistical Technique for Forecasting Concord Grape Production

Concord grape plot data consisting of number of clusters, number of berries per cluster, and cluster weight 60 and 90 days after bloom were combined with previous winter temperature data to develop objective forecasting equations for yield per hectare. Forecasting equations were developed for June, July, August, and September forecasts throughout the growing season. The mean absolute percent error for the forecasting equations decreased from 11.8% to 7.3% from June to September. The forecasting equations had lower mean percent absolute errors for the June, July, and August forecasts than the subjective forecasts made by the management of National Grape Cooperative.