Revisão sistemática de métodos computacionais baseados na biologia e métodos regressivos para a previsão de carbono total

Abstract

Over the years, the development of computational models has intensified. As a result, non-heuristic and heuristic models have emerged or changed. Many academic studies have been conducted and published on various topics using computational models as a basis for comparison. However, these studies tend to select only a few models based on the sheer number of models available. Instead of comparing all models, they instead compare a grouping chosen by each author. Therefore, in an attempt to broaden the comparisons made, this study aims to search for references in three consolidated databases (Scopus, Web of Science, and ScienceDirect) to compare the models used to solve the problem of predicting Total Organic Carbon (TOC). For this comparison, we analyzed a quality metric, R2, and three metrics that measure errors and deviations from the ideal or expected result: MSE, MAE, and RMSE. Using these metrics, several models used in TOC prediction studies were compared in an attempt to answer the question: ”Can the prediction of total organic carbon be obtained indirectly, through well log data, using methods based on natural or biological events, and achieve greater accuracy with a shorter execution time compared to machine learning methods based on regressive models?”