Machine Learning for predicting climate change in the environment: Review

Abstract

Climate changes currently occur abruptly and immediately being unpredictable by the population, causing damage and material losses, but with the support of current technologies, such as artificial intelligence: machine learning, will help us to anticipate these events. Therefore, this review aims to analyze the effectiveness of machine learning for the prediction of climate changes in the environment, to provide the validity of its performance and improvement. The methodology employed in this systematic review consisted of using PICO to establish eligibility criteria by grouping them into components that were finally reduced to PIOC, with which the following question was established, to what extent does Machine Learning improve the prediction of climate changes in the environment? which gave way to the development of the keywords for the creation of the search equation. Subsequently, the PRISMA methodology was used to discard articles by exclusion and inclusion, starting with a base of 2020 articles and after applying all the filters, 22 articles were included in the SLR. The results showed that machine learning showed superior performance in unraveling complex and interactive associations between environment and plant diversity, furthermore the ELM method generally provided superior accuracy to the other methods in predicting monthly soil temperatures at various depths. It was concluded that machine learning is an effective method that stands out among the other types of artificial intelligence showing a positive relationship to predict temperature changes in the environment, according to the approach presented, the most effective model that suits the research should be applied to obtain better results.