This work addresses the need for energy management in a microgrid (MG). Outliers in MG’s, representing anomalies or unexpected variations in data, can significantly disrupt the performance and reliability of energy management systems (EMS). The proposed work uses an unsupervised K-means clustering technique which is a prominent one for outlier detection and imputation. In addition to outliers, there is a need to model the uncertainties for effective EMS. This paper addresses the uncertainties based on Monte Carlo simulation for scenario generation and K-means clustering for scenario reduction. Furthermore, a two-stage scheduling approach is proposed to minimize day-ahead operational costs and real-time imbalance costs. The methodology was applied to an MG model, demonstrating reduced operational expenses and improved system stability by effectively managing outliers and optimizing scheduling decisions. The maximum energy exchange with the grid for a particular scheduling hour is limited to 70 kW. The total operational cost of the MG has reduced from 20,202.54 to 19,241.17 rupees, a reduction of 5% in the total operational cost of the MG.