A Statistical Approach to Forecasting State-Level Per Capita Electricity Consumption in India Using Linear Regression and ARIMA

Globally, electricity demand is rising amid
sustainability and reliability concerns. In India, state-level
planning is essential due to regional disparities in energy
consumption and infrastructure. Tamil Nadu, a major
industrial hub, faces evolving challenges in aligning generation
capacity with rising per capita consumption. Traditional
forecasting models like ARIMA are widely applied but often
yield poor accuracy with limited datasets or non-seasonal
patterns. This study addresses these limitations by applying
and comparing Linear Regression and ARIMA models to
Tamil Nadu’s power sector data (2006–2021), including
Installed Capacity, Generation, Purchases, and Per Capita
Consumption. Key metrics such as Purchase Dependency and
Generation Efficiency are derived, and Compound Annual
Growth Rate (CAGR) trends are analyzed. Forecasts through
2026 show that Linear Regression significantly outperforms
ARIMA in accuracy (R² = 0.93). The proposed approach
provides a reliable and interpretable framework for regional
electricity forecasting, offering policymakers a data-driven
foundation for future infrastructure and energy planning