Vehicle mobility is important to a diverse range of disciplines, e.g., geography, transportation, and public health. Machine Learning algorithms have been applied in geospatial analysis related to vehicle mobility and travel pattern research, which provided researchers with more flexibility and capabilities for complex mobility pattern analyses. This dissertation aims to explore how different Machine Learning models (e.g., regression and clustering) can be applied to enhance the interpretability of vehicle mobility patterns by conducting explanatory analyses on factors that may impact different mobility patterns (i.e., trip volume changes and travel times) over space and time (e.g., different stages of the COVID-19 Pandemic at regional and nationwide scales). In this dissertation, three studies were undertaken to investigate the spatiotemporal trends of vehicle trip changes and travel behaviors, using passively-collected mobile device data. The first study examined mobility patterns over different time periods during the summer 2020 when COVID-19 cases were spiking in Florida(locations with large numbers of vulnerable individuals) and analyzed a set of underlying drivers for mobility and how these factors changed over time using Machine Learning approaches. The second study investigated changing mobility patterns across the U.S. during 2021 when COVID-19 vaccinations were becoming available to understand whether changing vaccination rates led to a change in the rate of trips using Machine Learning clustering methods. The third study investigated reasons impacting travel times for two origin-destination pairs using a Machine Learning approach to better understand how different factors can affect travel times over different trip purposes and different trip lengths in Maryland. The contributions of this dissertation are that it provided new insights into how different types of mobility patterns evolved over space and time, especially during a major public health crisis, and the results are useful for policy and planning implications for local and regional officials, e.g., mobility restriction measurements, decision support for economic recovery, and public health strategies. The integration of diverse data sources (e.g., passively-collected mobility data and other mobility data from different public and private sources) and the utilization of multiple Machine Learning models enhanced the interpretability of vehicle mobility patterns.