Nowadays, various design techniques are proposed for implementing different kinds of solution for the software desired for complex systems. Among them design patterns are frequently used to identify a better solution for the object-oriented software. The recognition of design pattern in open source program enables the maintainability of traditional object-oriented programs. The proposed methodology shows the determination of design patterns in open source program. The process of pattern determination is done by using quality measurement parameters e.g., software metrics. The presented method considers various kind of neural network models to determine the presence of design patterns in open source system. The proposed methodology is evaluated by using a case study e.g., JUnit. Subsequently, a comparative analysis is presented to identify the strengths and weaknesses of selected neural network models. © 2019 IEEE.