Data-driven time series demand forecasting of Car-Sharing services: The case study of DriveNow in Munich, Germany

Sharing-vehicle systems are one of the tools that municipalities are trying to apply in order to resolve environmental and traffic issues. The trend of the usage of car- sharing is increasing with high rate of new subscribers each year. Therefore, better performed operations in car-sharing systems lead to a higher acceptance as mode of transport from inhabitants of urban areas. This research proposes the methodology to apply Machine Learning technique such Neural Networks to forecast the Demand of car-sharing vehicles in the case study of Munich. The study analyses the figures and trends of the share-mobility in Munich and investigate on the Business of the DriveNow in Munich. The thesis reveals that major influences areas within the city (e.g. the Airport) and applies a reclassification of the areas based on district level in order ease the computation of the model. The results of the Neural Network are compared with traditional ARIMA time-series forecasting describing pros and cons of both techniques.     «

Sharing-vehicle systems are one of the tools that municipalities are trying to apply in order to resolve environmental and traffic issues. The trend of the usage of car- sharing is increasing with high rate of new subscribers each year. Therefore, better performed operations in car-sharing systems lead to a higher acceptance as mode of transport from inhabitants of urban areas. This research proposes the methodology to apply Machine Learning technique such Neural Networks to forecast the Demand...     »