Predicting peak demand for electricity consumption using time series data and machine learning model

Aug 5, 2026

Energy consumption is driven by factors such as the spread of electronic devices, technological progress, economic growth, agricultural development, and population increase. This study targets the prediction of peak energy demand (ED) using historical five-year time series data together with temperature data from official Tamil Nadu sources. Feature engineering prepares inputs for machine learning models including XGBoost regressor, lasso, and ridge regression, and time series models including ARIMA and VAR. Results indicate accurate ED forecasting to support energy planning and resource management decisions.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages9

Reading Time23 min

AccessFree to download

Introduction

Background and problem motivation

Existing approaches and limitations

Proposed approach and data integration

Indonesian Journal of Electrical Engineering and Computer Science Vol. 38, No. 1, April 2025, pp. 668~ 676 ISSN: 2502-4752, DOI: 10. 11591/ijeecs.v38 . i1 .pp668-676 􀂈 668 | Predicting peak demand for electricity consumption using time series data and machine learning model
Suriya S., Agusthiyar R.
Department of Computer Science and Applications, SRM Institute of Science and Technology, Chennai, India | | | --- | --- | | Article Info
Article history:
Received Jul 12, 2024 Revised Oct 3, 2024 Accepted Nov 10, 2024
Keywords:
ARIMA Lasso Multivariate Ridge Univariate XGR regressor
Corresponding Author: | ABSTRACT
Energy consumption is influenced by various factors, including the proliferation of electronic devices, technological advancements, economic growth, agricultural development, and population increase. Each of these factors contributes to the rising demand for energy. This paper addresses the challenge of predicting peak energy demand (ED) by utilizing historical time series data from the past five years, combined with temperature data from Tamil Nadu’s official sources. We employed feature engineering techniques to prepare the data for machine learning models, specifically XGBoost regressor, lasso, and ridge regression. The time series data was then analyzed using both univariate and multivariate models, including auto regressive integrated moving average (ARIMA) and vector autoregressive (VAR) models. The results show that our models can effectively forecast ED, providing critical insights for policymakers and stakeholders involved in energy planning and resource management.
This is an open access article under the CC BY-SA license.
| | Suriya S.
Department of Computer Science and Applications, SRM Institute of Science and Technology Chennai, India
Email: [suriyas@srmist.edu.in](suriyas@srmist.edu.in) | | 1. INTRODUCTION Accurate forecasting of peak energy demand (ED) is crucial for effective power resource management and infrastructure planning. In Tamil Nadu, like many other regions, ED fluctuations can significantly impact the efficiency of power supply systems. Inaccurate predictions can lead to over or underutilization of resources, affecting both operational costs and energy reliability. The challenge lies in developing predictive models that can accurately capture complex patterns and trends in energy consumption. Existing solutions for forecasting peak ED typically involve time series models and machine learning techniques. Univariate models, such as the auto-regressive (AR) integrated moving average (ARIMA) model, are commonly used due to their simplicity and ability to capture historical patterns. However, these models often struggle with irregularities and complex patterns, such as those induced by economic changes, policy impacts, or sudden disruptions. On the other hand, multivariate models like the vector autoregressive (VAR) model offer a more holistic approach by incorporating multiple variables, but they can be computationally intensive and require careful parameter tuning. The major constraints in forecasting peak demand include data quality issues, such as missing values and inconsistencies, and the challenge of capturing external factors that influence energy consumption. One of the critical aspects of energy management is the prediction of peak ED, i.e., the maximum amount of energy required at any given time [1]-[7]. Accurate forecasting of peak demand enables utilities to optimize resource allocation, mitigate supply-demand imbalances, and avoid potential grid failures or blackouts. However, predicting peak demand is a complex task influenced by a myriad of factors, including weather patterns, socio-economic trends, and technological advancements. In this paper, we address the challenge of predicting peak ED in Tamil Nadu by leveraging advanced data analytics techniques, specifically machine learning and time series analysis [8]-[16] . Our appr

FAQs

What data sources are used to predict peak energy demand in Tamil Nadu?

The method combines historical five-year energy consumption time series with temperature data from official Tamil Nadu sources.

Which machine learning and time series models are applied?

Machine learning includes XGBoost regressor, lasso, and ridge regression, while time series analysis uses ARIMA and vector autoregressive (VAR) models.

Why is predicting peak energy demand important?

Accurate forecasting helps optimize resource allocation, reduce supply-demand imbalances, and improve energy reliability by enabling better infrastructure planning.

Aug 5, 2026

Upload by Evangeline