Electrical Power Consumption Analysis using Linear Regression Analysis and Rough Set Theory

The advanced and evolving features of fast internet-of-things (IOT) based power consumption monitoring system for households and buildings are critical to the success of monitoring and logging the consumption behavior of the consumers. This is necessary to understand and calculate the electrical consumption based on real-time power measurements and help in identifying effective energy conservation and efficiency measures. In this paper, linear regression analysis and rough set theory (RST) simulation were used to determine the correlation of school building operations parameters, such as number of school days per month, and student populations in its monthly power consumption. While linear regression was able to provide the correlation among these factors, RST has provided a more detailed perspective on the consumption behavior and category according to its pre-defined attributes or conditions. RST analysis reveals a higher classification accuracy when leaving-one-out validation method and combination of strength and similarity rule support is applied. Likewise, the increase in dataset has significantly improved, yet has resulted in a different core reduction results, and decision-making process from the available attribute-value pairs.