The Internet of Things (IoT) brought the idea of smart cities to the world, spurred by the rapidly growing number of linked products. The integration of several aspects into smart city architecture, including smart transit, smart community, smart health care, smart grid, smart technology, smart service, etc., is to improve the standard of living for urban people. Real-time data processing needs and exponential data management, however, prevent smart city realisation. The BDA-embedded smart city serves two main purposes. In order to plan, construct, and operate smart cities, it first makes it easier to use urban big data (UBD). In order to improve the calibre of urban services, BDA is also responsible for managing and processing large amounts of UBD. Utilising LBFMF to incorporate data normalisation and data filtering techniques into the suggested task. We were able to obtain the output values needed for managing municipal operations and urban planning by analysing the data sets. The aforementioned data sets are used to derive performance metrics for the proposed BDA with ML approach in terms of correlation is 0.96, MSE is 43.41, PSNR is 38.72, and time of execution is 12.72 ms for data processing. An investigation of productivity and processing time with respect to prior works guarantees the performance superiority of the suggested job. Therefore, we can affirm that the suggested BDA + ML implementation is reliable and applicable for developing urban smart city architecture in the real world.