Predictive Service Allocation for Informal Automotive Mechanics Using Machine Learning: A Case Study from the Big Five Startup

The informal automotive sector in South Africa is crucial for its contribution to economic growth and job creation. Due to it being informal, it is largely excluded from the mainstream of formal businesses, resulting in less benefit from what larger formal businesses benefit from. In trying to understand how service delivery might be improved in South Africa’s informal automotive sector, this project looked at the possible role of digital technologies, particularly machine learning. The main idea was to see whether a simple prediction model could help manage common delays in service response, especially in areas where informal mechanics operate without structured systems. For this purpose, we designed a basic model and tested it on the Big Five platform. Since actual service data were not available yet, we relied on a simulated dataset to train the model. The development process used PyTorch, which allowed for flexibility in adjusting the architecture during early testing. The dataset itself was designed to reflect real-world conditions as closely as possible, incorporating factors like the type of mechanical problem, where the request came from, whether urban, township, highway, or rural, what kind of vehicle was involved, and what sort of service was being requested. Although based on synthetic data, the model achieved promising initial results. It reached a mean squared error of 3.8 min2, a mean absolute error of 1.6 min, and an R2 value of 0.82, indicating reasonably strong predictive performance under test conditions. But beyond the numbers, this study sheds light on how predictive tools can play a key role in promoting digital inclusion, especially in informal markets that have long been excluded from structured service and supply chains. The model demonstrates how Artificial Intelligence can help small-scale mechanics move towards more formalized operations, improving both customer satisfaction and operational performance.