Globally, healthcare administrators are working to reduce healthcare costs without sacrificing the quality of service provided. The biggest portion of health expenses is related to hospitalization. Length of stay has a direct impact on the hospitalization cost and resource management. Researchers for the last few years are trying to find the relevant factors to accurately identify LOS. In this research, a strategy to forecast the number of hospitalization days in a community was developed using USA SPARCS 2017 de-identified data repository. The data is not individually identifiable. The dataset consists of 2,343,569 instances and 34 features. In this study, we are proposing a three stage model: admission, post admission, and discharge stage for an improved hospital length of stay prediction. These stage-wise models will help both the patient and hospital management to correctly predict LOS from the beginning of the hospitalization to the time of discharge. This study suggests a paradigm for predicting patient length of stay (LOS) using various machine learning (ML) and ensemble techniques. This stage-wise prediction will be performed using a weighted probability averaging ensemble model. The weights of the individual classifiers will be selected based on their performance through a grid search method. The proposed ensemble model has outperformed the individual base ML models. The proposed ensemble model has improved balanced classification performance and it improves across stages: admission to discharge. The discharge stage has the highest accuracy 0.9065 followed by the post admission stage with 0.8921 and the admission stage has the lowest among all 0.6824.