Building transparent knowledge-based systems in the form of accurate and interpretable fuzzy rules is one of the significant applications of fuzzy set theory. The fuzzy connectives, i.e., T -norm/conorm, play the role of connecting fuzzy sets, which are essentially linguistic terms extracted from the knowledge embedded in a given data set. Fuzzy pattern tree is a recently proposed novel machine learning technique, which grows a hierarchical binary tree for each known class utilising conventional T -norms/conorms and aggregation operators. Preaggregation functions are recently proposed in the literature as a type of generalised aggregation functions, which have achieved successes in a number of applications. This paper proposes a preaggregation-based approach with application to the construction of fuzzy pattern tree. An experimental study is done to explore the performance of the fuzzy pattern tree where preaggregation functions are employed in comparison to that where conventional aggregation operators are utilised. Experimental results demonstrate that the performance of fuzzy pattern tree incorporated with the preaggregation function generated by Nilpotent minimum T -norm outperforms those with alternative preaggregation functions and the commonly used ordered weighted averaging operators.