Analysing Electricity Consumption in Malaysia via Linear Programming, Time Series Analysis and Clustering Analysis

Abstract

Malaysia faces increasing electricity consumption that later creates challenges in long-term sustainability. The lack of accurate forecasting and clear consumption pattern classification limits effective energy planning. This study aims to minimize electricity transmission losses using linear programming, forecast future electricity consumption using time series analysis and categorize consumption levels using clustering analysis. Monthly electricity consumption data from January 2018 to May 2025 were obtained from the Department of Statistics Malaysia (DOSM) and analysed using Python. In the first phase, linear programming was applied to optimize electricity distribution among industrial, domestic and export sectors while satisfying supply and demand constraints. The results showed that optimized allocation reduced annual electricity losses by 60 to 80 MKWh that proves the model's effectiveness. In the second phase, time series analysis such as Seasonal Autoregressive Integrated Moving Average (SARIMA) model was developed to forecast future monthly electricity consumption. Mean Absolute Percentage Error (MAPE) showed the high predictive accuracy below 2% which indicates a stable increasing consumption trend. In the third phase, clustering analysis such as K-means clustering and Principal Component Analysis (PCA) were used to categorize monthly consumption levels. Results showed upward shift toward higher consumption after 2023 because of industrial growth and rising domestic demand. Overall, linear programming, time series analysis and clustering analysis successfully provide a comprehensive data-driven framework for sustainable electricity management. The findings offer valuable insights for policymakers and the Energy Commission in formulating strategies to reduce losses, enhance forecasting accuracy and improve energy distribution efficiency across Malaysia’s electricity sector.