Conventional power grid which allows single directional power flow is basically controlled and managed using electromechanical systems. Such conventional system has many problems including high loss, no responsiveness to changes and so on. To solve these problems smart grids have been proposed. Smart grid is a Bi-directional electric grids and communication networks that improve the reliability, security, and efficiency of the electric system for small- to largescale generation, transmission, distribution, and consumption. Smart grid deployment is a global trend, creating endless possibilities for the use of data generated by dynamic networks. Smart Grid technologies create greater volumes of data. The challenge is the transformation of this large volume of data into useful information for the electrical system. An example is the use of smart meters. A Smart Grid monitors electricity delivery and tracks power consumption with smart meters that transmit energy usage information to utilities via communication networks. Smart meters allow for two-way electronic communication in which valuable information flows dynamically between consumers and electricity producers. This large data can be effectively sent for load forecasting. To this end, this thesis deals with the optimal use of data collected using smart grid for load forecasting. Neural network based optimal load forecasting for improved demand response has been studied. The result shows that load forecasting using neural network enhance power system management and operation with minimum error. Moreover, the application of demand side management (DSM) or demand response (DR) techniques for the optimization of power system management in real time has been analyzed.Furthermore, the importance of introducing smart meters in Ethiopia and deployment strategies has been addressed. For the analysis, Mat lab software has been utilized. In addition, the status of smart metering in various countries is also illustrated.