A Prediction Approach for Demand Analysis of Energy Consumption Using K-Nearest Neighbor in Residential Buildings

In order to manage efficiently the energy production, storage and management system, it is very important to analyze accurately the energy requirements for residential sector because the residential sector consumes a considerable amount of total energy produced. The main aim of the paper is the assurance of energy production according to the consumer demands in an efficient manner. The energy market is an important tool for setting prices between the energy producers, suppliers and the consumers. An excellent precision in the prediction of next day consumption is required for the suppliers to get good prices in the energy traded. The main aim of this paper is to facilitate the energy suppliers to make decisions for the provision of energy to different apartments according to their demand. In this paper, we have utilized K-Nearest Neighbors classifier for daily energy consumption prediction based on classification. The process consists of five stages namely data collection, data processing, prediction, and validation and performance evaluation. The historical data containing hourly consumption of 520 apartments of Seoul, Republic of Korea has been used in the experimentation. The data has been divided into different training and testing ratios and different qualitative and quantitative measures have been applied to find the performance and efficiency of the predictor. The highest accuracy has been observed for 60-40% training and testing ratio giving 95.9615% accurate results. The effectiveness of the model has been validated using 10Fold and 5-Fold cross validation.