Predicting Demand in IoT Enabled Service Stations

The current world of AI revolves around forecasting and prediction, eating up a major chunk of problem statements generated by organizations in various domains. Applying forecasting techniques to get an advance view of the number of visitors per bay in a service station will help the owners to plan and adjust their resources for increased operational efficiency as well as to manage any sudden change in visitor demand. As part of this paper, we create a prediction model for forecasting expected visitors demand for a service station. We compare the results of applying various machine learning and statistical methods using historical visitor data as well as historical weather data. The result states that, among the techniques used, neural network using both historical and weather data performs best, providing a future view of expected demand with high accuracy.