Frequent pattern mining has attracted much attention and wide applications owing to its simple concept and strategy. It is of the most important task in data mining and knowledge discovery. But usually a large number of frequent patterns get generated from a large scale of data matrix which is a time consuming affair. So, in order to discretize the data matrix a mathematical concept called fuzzy logic was used. It generalizes the data matrix values in the range of 0 to 1. In the due course of time, an evolutionary algorithm, called Particle swarm optimization (PSO) has also gained much popularity. But due to the premature convergence of PSO, a comprehensive learning strategy was introduced that used all particles’ best information to update a particle's velocity. It also enabled the diversity of the swarm to be preserved to discourage premature convergence. In this paper, frequent patterns were generated from the fuzzy dataset (data matrix converted into fuzzy data matrix) using the Frequent Pattern (FP) growth algorithm. In order to generate some of the best individual frequent patterns out of the entire set of patterns, the CLPSO algorithm was used with a selection measure called mean squared residue (MSR) score. It was noted that the CLPSO algorithm outperformed the traditional PSO algorithm in the generation of the best individual patterns with a comparatively lower MSR value.