AbstractThe study's objectives are to factorize industrial sector carbon emissions (ISCE) into eleven predetermined factors, predict ISCE and validate the predicted ISCE value. The study uses multiplicative Logarithmic Mean Divisia Index (LMDI), Artificial Neural Network (ANN) and Random Forest (RF) models to examine ISCE between 2000 and 2020. According to the multiplicative LMDI, the strongest inhibitors of ISCE were Sectoral Energy Intensity (75 percent), Natural Resource Productivity (60 percent), and Energy Tech Research Investment Productivity (57 percent). The top enhancers of ISCE were Natural Resource Consumption (81 percent), Overall Research Priority (80 percent), Carbon Emission Factor (67 percent), and Energy Research Efficiency (60 percent). This suggests that strategies for reducing ISCE may be most successful when they focus on these four major enhancers. The Random Forest algorithm performs better than Artificial Neural Network in terms of prediction, as evidenced by its stronger linear connection, better fit, and lowest RMSE. However, the difference in performance is not much (calculate performance from RMSE values). With these values, data envelopment analysis can be used to determine the possible reduction in ISCE, which indicates a direction for further study.