Abstract:
Electricity consumption has progressively increased all over the world, including Thailand. Thus, the research that studies the affected factors and the predictions on electricity consumption is mandatory. This research collects electricity consumption data from Kasetsart and Chulalongkorn universities between 2013-2018. The university buildings are classified into 6 categories: lecture halls, administrations, multi-purpose buildings, research laboratories, dormitories and parkings. This research analyzes affected factors using feature selection. Furthermore, machine learning is applied to create a predictive model based on interval data for electricity consumption, using multiple regression, neural network and random forest. The result shows the overall affected factors are area and number of stories. Different categories have different affected factors. Moreover, random forest provides more accurate consumption prediction, compared to the other two methods. The results can also be applied with other buildings to analyze electricity consumption prediction.
BibliograpyCitation :
ใน King Mongkut's University of Technology North Bangkok Faculty of Information Technology and Digital Innovation. The 16th National Conference on Computing and Information Technology (NCCIT 2020) (p.584-589). กรุงเทพฯ : คณะเทคโนโลยีสารสนเทศและนวัตกรรมดิจิทัล มหาวิทยาลัยเทคโนโลยีพระจอมเกล้าพระนครเหนือ