Predicting the impacts of key development indices on the ecological footprint in Afghanistan using deep learning

Abstract Evaluating the ecological footprint (EF) is one of the objectives of nations worldwide, playing a vital role in preserving their environmental resources. This check aims to predict the impacts of key development indices on the EF using deep learning methods with time series data for the period of 1980–2019 in Afghanistan. Initially, an auto-encoder neural network test was used for the analysis of the time series data. The dataset was split into a training set comprising seventy percent of the data and a test set comprising thirty percent. Secondly, auto-encoder neural network methodologies have attracted substantial attention due to their deep learning capacities, offering data optimization and enhancing the accuracy and precision of predictions in both dependent and independent variables. Thirdly, the reliability, stability, and predictive capabilities of the parameters were assessed using an auto-encoder neural network through preliminary tests. The results of the diagnostic tests confirm the predictability and reliability of the parameters in the auto-encoder neural network model. Notably, a strong positive relationship is observed among development indices and EF. The highest correlation coefficient is observed between the total population index and the EF, yielding a rate of R = 0.94. Furthermore, a correlation coefficient of 0.91 is found between the agricultural production index and the ecological footprint. Therefore, on these findings, it can be inferred that the development indices exert significant positive effects on the EF in Afghanistan. To conclude, this study showed deep learning methods can be utilized to predict the impact of development indices on the EF in Afghanistan.