<p>Fuzzy logic (FL) offers significant capabilities in multi-classification tasks as it handles imprecise and uncertain data for nuanced decision-making. However, the development of precise fuzzy sets and rules requires time and considerable effort and expertise. In fact, the computation time increases with the number of rules due to combinatorial complexity. For a robust FL model, effective data description, knowledge extraction, and representation for rule induction are essential. To tackle the mentioned challenges, this work extends the Integrated Truth Table in Decision Tree-based Fuzzy Logic model (ITTDTFL) to the Integrated Truth Table in Tree-based Fuzzy Logic model (ITTTFL). The latter incorporates PCA and WPD for feature engineering and builds the best-fit fuzzy sets and optimized rules based on Tree-based Machine Learning algorithms. More specifically, the model uses the C4.5 algorithm and Random Forest-based decision trees to extract rules and membership functions and generate optimized fuzzy rules using a Truth Table (TT). ITTTFL model extensions are compared with state-of-the-art models, including XGBoost, Random Forest, C4.5 Algorithm, FURIA and RIPPER on three real datasets from a pump and UCI datasets. The experiments consists in evaluating performance based on factors such as the number of rules generated, accuracy, and computation time. The experimental results show that our proposed models performed better, achieved excellent accuracy with a minimum computation time, outperforming the other models.</p>