Obtaining the Demand Curves of Commercial and Industrial Customers of the Energy Distribution Company EEASA through Telemetry and Data Mining

This article presents the extraction of demand characteristic curves for commercial and industrial customers within the concession area of EEASA (Electric Company Ambato). Machine learning classification algorithms are employed to process telemetry data from 184 customers. The methodology is divided into four stages: firstly, consumption data is downloaded and stored in Excel. Then, the data is cleaned, sorted, and classified, and the k-means algorithm is applied to create a data matrix processed in Python. In the third stage, Principal Component Analysis (PCA) is used to reduce data volume and characterize seasonal patterns. Finally, the Fuzzy-c-means algorithm is applied to obtain a seasonal curve that serves as a reference for user behavior. As a result, demand curves that characterize consumers are obtained. These findings are of great importance for the management and planning of the electric company as they provide detailed information about the consumption patterns of their customers, enabling better resource allocation and more efficient response to energy demands. Additionally, the use of machine learning algorithms demonstrates their effectiveness in processing and analyzing large volumes of data, which can be applied in other contexts and industrial sectors.