The distributed generation in smart grids has become highly prevalent due to its inherent characteristics like robust, reachable, lossless and emission less. The increased usage of sensor devices, wireless and network communication, cloud computing and IoT (Internet of things) explores the merge of smartness in energy field which results in an extensive collection of data getting populated in the electricity sector. This paper proposes a hybrid machine learning with big data analytic techniques for optimized distribution of available energy resources targeting smart energy management. The system aims in handling smart energy management using the data received from conventional sources and various distributed generation sources like photovoltaic, wind, small hydro and biomass. For unsupervised power data an efficient clustering methodology such as grid based clustering has been incorporated for optimal distribution of energy received from the nodal and zone regions. For structured power data, support vector machine algorithm of supervised learning is used for smart distribution which performs classification and regression of the vast electricity data. Regressive analysis over electricity data across the country with respect to various regions is carried out to understand the energy consumption which gives the insight of energy deficit. Thus the machine learning hybrid techniques were performed over the collected data to validate the smart distribution and the result obtained ensures a substantial gain which leads to conservation of generated energy through smart energy management.