With buses being a primary mode of transport for the common man, it is imperative that the bus transportation system works with optimal efficiency with respect to its resources. In Mumbai, there are 3600 buses running over 422 routes, managed by Brihanmumbai Electric Supply and Transport (BEST) and carrying millions of citizens every day. The fleet distribution is currently updated every four months manually by BEST officers. As a result, scheduling and allocating solution of BEST is time consuming, error prone and inconsistent. The shortcomings of current approaches include the highly dynamic nature of traffic, consideration of infinite bus capacity and redefining routes completely. As the variation in the data over the short term weekly cycles is difficult to fit using linear and polynomial regression models, we find it unsuitable for our problem. Other regression models suffer from the lack of availability of attribute values for the next day to generate trip forecasts. Hence, we propose a solution to dynamically allocate buses to routes and generate their running schedules. We considered ARIMA and SARIMAX models that allow forecasting trips that are implemented using real-world data. Based on the forecasted data, fleet size allocation is done. The results show a 6.52% decrease in buses required using ARIMAX model and a 3.26% decrease in resources using SARIMAX. These models also maintain the average headway required 80% of the time.