Utilising tomato seed sales data for strategic decision-making in a South African firm: A forecasting approach

Abstract

The mini-dissertation examines using tomato seed sales data as a forecasting tool to enhance strategic management decision-making within a South African seed company. This research is underpinned by four hypotheses, focusing on the influence of price, production costs, production volume, and weather conditions on tomato seed sales. Using Random Forest Regression (RFR) as the primary forecasting model, this study leverages historical data from 2013 to 2022 and incorporates climatic variables to improve the predictive capabilities of sales forecasts. The data-driven approach is supported by a comprehensive literature review covering both traditional forecasting methods and the application of Machine Learning (ML) in agricultural forecasting. The research also employs descriptive statistics, correlation analyses, and regression models to evaluate the relationships between production factors and sales outcomes. Key findings indicate that all four variables significantly impact seed sales, with weather conditions and production costs playing particularly crucial roles in shaping sales performance. The forecasting model's results suggest that improved accuracy in predicting these variables can enhance the company's ability to plan production and manage inventory, leading to better cost management and increased profitability. The managerial implications of this study include recommendations for differentiated pricing strategies, investment in climate-resilient technologies, and adopting predictive analytics to align production with market demand better. The study also highlights the need for expanding geographic coverage to improve yield optimisation and reduce production risks. Limitations include the reliance on secondary data and the scope restricted to specific geographic regions, which may limit the generalisability of the findings. This research contributes to the broader field of agricultural sales forecasting by demonstrating the effectiveness of ML models, particularly RFR, in improving the accuracy of demand predictions in a complex and dynamic market. The insights provided are intended to support seed companies in making informed, strategic decisions to ensure profitability and sustainability in a highly competitive industry.