A machine learning framework for predicting Bangladesh's economic growth: Emphasizing the sector-specific carbon emission dynamics

Accurate prediction of gross domestic product (GDP) is essential for informed policymaking, economic planning, and sustainable development strategies. While many studies had used socio- and macroeconomic factors to predict the GDP of various nations, few had explored sector-specific CO 2 emissions as a predictor, despite their well-established relationship with economic growth. This study addressed this gap by utilizing sector-wise CO 2 emissions data of Bangladesh from 1970 to 2022 to forecast GDP and evaluate the performance of four machine learning algorithms: random forest, support vector machine, gradient boosting machine, and extreme gradient boosting to predict GDP. The performance of these models was evaluated using root mean squared error, mean absolute error, mean absolute percentage error, coefficient of determination, and percent bias metrics. Historical data showed that the power generation sector experienced the most substantial rise in CO 2 emissions throughout the period, followed by industrial processes, transportation, buildings, and industrial combustion. A slight decrease was observed in 2020 as a result of the COVID-19 pandemic. Among the models, the random forest showed better performance. Residual analysis and the percentage of relative error index further confirmed the random forest's reliability, as it showed minimal deviations between actual and predicted GDP values. Feature importance analysis highlighted that CO 2 emissions from industrial combustion and transportation sectors had the most significant impact on GDP, followed by the power and building sectors, while agricultural emissions had minimal effect. These findings suggested that sector-specific CO 2 emissions could effectively predict GDP fluctuations, offering valuable insights to policymakers for balancing both economic growth and environmental sustainability. • The study predicted GDP using sector-specific CO 2 emissions in Bangladesh (1970–2022). • Four ML models (RF, SVM, GBM, XGB) were evaluated for GDP prediction performance. • All models performed well, with RF showing the best performance. • CO 2 emissions from industrial combustion and transportation had the most significant impact on GDP. • The findings provide insights into balancing economic growth and sustainability.