Logo image
Forecasting grape production based on vineyard characteristics and management practices in South African wine regions
Thesis   Open access

Forecasting grape production based on vineyard characteristics and management practices in South African wine regions

Lekau Dan Madisha Madisha
Master of Applied Data Science, University of Johannesburg
2025
Handle:
https://hdl.handle.net/10210/521240

Abstract

This study examined vineyard and grape production data from 2015 to 2024 across four South African regions: Northern Cape, Olifants River, Paarl, and Cape South Coast. The integrated dataset included over 570,000 records covering 81 cultivars. An analysis of production trends revealed a dip in 2018, a peak in 2021, and a gradual decline through 2024. The panel regression models showed that total grape tonnage was positively associated with vineyard area and negatively associated with vineyard age. The yield per hectare varied significantly by irrigation type and trellis system. Drip irrigation generally produced higher yields than flood irrigation, while trellis systems such as T trellis and two strand hedge outperformed bush vine and Perold configurations. These effects remained robust after controlling for region, cultivar, vineyard size, and age. Forecasting with a walk-forward ARIMAX model incorporating structural and management variables outperformed baseline methods, demonstrating strong predictive accuracy and well-calibrated confidence intervals. The findings highlight the importance of vineyard structure and management practices in influencing annual grape yields.
pdf
Madisha LD- mini dissertation983.97 kBDownloadView
Open Access

Metrics

2 File views/ downloads
2 Record Views

Details

Logo image