Exploring micromobility dynamics through Machine Learning prediction algorithms: an analysis of urban transportation patterns

The term "micromobility" has just recently entered our lexicon (approximately 2017) and it can be defined simply as mobility pertaining to short routes and distances, primarily in cities. Micro vehicles, which have a light mass and a constrained speed, are included in the idea of micromobility, both powered and unpowered, private and shared vehicles are covered in this list. The shared use of vehicles, specifically bicycles, is the main topic of this thesis. Micromobility sharing services have expanded significantly since this idea was first proposed in many parts of the world. In this study, the usage of these services was analyzed using some Machine Learning models such as: ARIMA, Linear Regression, Lasso, Ridge, Random Forest and Gradient Boosting. The goal of this thesis is addressing the problem of predicting the availability of bikes for a station-based sharing service and the flux of bikes in a certain area. Specifically, it investigates the applicability of machine learning models to forecast the number of occupied slots of a particular sharing station of bikes in San Francisco (USA) and the number of bikes that cross the Fremont Bridge in Seattle (USA). To accomplish this, a methodology has been developed. First, the data was collected, after which it was cleaned, integrated, and subjected to preliminary analysis to determine how best to manage it. Next, the chosen machine learning models were trained using the sliding window technique, their performance was compared, and the best machine learning model was chosen. Lastly, by analyzing temporal and weather contexts, using the chosen Machine Learning model trained on past data, we obtained a good future prediction on bike occupancy. The outcomes allow us to provide practical guidelines to setup and tune Machine Learning models on these fields. As a practical result of this work, it is possible to forecast the arrival of bikes or, more broadly, people in a certain urban area, which might help businesses or local governments to enhance the services they provide there.