In recent years, we have witnessed the rise of brute force methods in the development of artificial intelligence applications. For decades, efforts in artificial intelligence have sought to emulate human reasoning to create “intelligent” reasoning models akin to human cognition. However, experts often struggle to identify the mental rules guiding their decisions, and not all the rules they believe they use are reliable. Conversely, testing all possible factor combinations to find correlations between concepts can be inefficient and may result in testing numerous absurd combinations. In this paper, we present a prototype proposal aimed at improving the accuracy of fuzzy modeling in two key aspects by analyzing available datasets. Firstly, it enhances the truth values of the definition of fuzzy concepts based on the distribution of available data. Secondly, it provides a more accurate credibility assessment for defined fuzzy rules according to their success with respect to previous datasets. Both improvements contribute to achieve more reliable and accurate predictions.