The continuous increase in urban population and the long-term prevalence of car dependency have led to a series of traffic issues such as traffic congestion, accidents, and environment pollution. One response to these problems is to enhance public transportation (PT). PT serves as a vital mode of mobility for residents, offering relief to urban traffic problems. To encourage and develop PT effectively, a profound understanding of its associated travel behaviours and operational efficiency is essential. Traditionally, this understanding has primarily relied on conventional survey methodologies. In recent years, a wealth of travel record data has been generated within the PT domain, including but not limited to smart card data, GTFS data, and automatic vehicle location data, with the advancement of information technology systems. Compared to manually collected passenger survey data, these digital datasets offer advantages in terms of accuracy and scale, presenting new opportunities for analysing and comprehending PT travel characteristics and operation performance. However, such data have not been fully exploited, In this context, it is necessary to make full use of massive multi-source PT data to make up for the shortcomings of nonaggregate data sources and develop a reasonable data-driven model to analyse the characteristics of PT passengers travel behaviour and the operating performance, which can provide the basis for further decision-making and optimization management plan to better meet the travel needs of PT passengers and improve service level of PT. Aiming to understand when, where and how passengers travel with public transportation in Southeast Queensland, this study proposed a data-driven framework to investigate and analyse the public transit travel characteristics, especially for bus travel in Brisbane. First, the study processed multi-source heterogeneous data, encompassing data cleaning, restructuring, fusion, statistical analysis, and computation among other procedures. Spatial and temporal travel characteristics were extracted, such as stop throughput, travel time and transfer time. Combined with Geographic location information for data visualization, revealing the travel patterns of PT passengers. it is clearly demonstrated that there are obvious variations in passenger flow at different stops by extracting the heat map of passengers boarding at a specific bus station during a certain period. Then a comprehensive investigation of public travel characteristics was performed. It can be found that the PT travel patterns exhibit significant differences across temporal and spatial dimensions between weekdays and weekends, as well as among different stops and routes. This data-driven analysis framework helped us to have an accurate and holistic understanding of PT usage and passenger travel activities. In addition, methodologically, this research developed a two-tier framework, which consisted of a criteria tier and an indicator tier, to evaluate and analyse the PT operation performance based on multi-source data. Four criteria consisting of various indicators were selected and weighted for a case study on a public bus system in Brisbane, Australia, which was carried out to demonstrate the usefulness of the framework. Multi-source data such as smart card data, GTFS data, household travel survey data and geographical data were cleaned and fused to obtain effective features. The operational performance of the public bus system was analysed and assessed on three layers: the indicator layer, the criterion layer and the system layer. The results demonstrated that the framework could identify the deficiencies of the PT services and provide the basis for further operating performance enhancement. Using Southeast Queensland as a case, this study examined the spatial-temporal characteristics of PT travel behaviour and the operating efficiency of bus services using extensive multi-source data, the findings provided a reasonable foundation for urban transportation management planning and decision-making. Methodological and theoretical contributions are also rendered in this research, including (1) the methods to process, combine and mine multi-source data for PT-related research (2) the proposed data-driven framework for comprehensive analysis of PT travel characteristics (3) the proposed data-driven framework to assess and analyse the operating efficiency of PT.