1. Business Understanding As part of datathon 2020, our case is tasked to build a predictive model to public transport services. The model can detect when the machine plan for less disruption in the wake of severe weather conditions and leverage the emergency management plan as well as providing uninterrupted services and products to citizens. […]
The purpose of the project was to build a model that tries to predict potential delays in Dubai’s bus transportation schedule, based on the weather conditions. Additional Extreme Gradient Boosting model was built, which is based on the weather conditions by 5 hours ago, which slightly improved the prediction of a few outliers, although this came at the cost of reducing the prediction accuracy for non-outliers. The overall prediction power was unfortunately unimpressive and could potentially be improved by analyzing the bus transportation data at an hourly level, by including additional data, such as global weather forecasts and traffic estimates, but also by exploring more feature engineering options, for example – seasonality, business activity, hourly segments and outlying flags.