Enhanced Model for Mining Software Repositories

This study presented Enhanced Model for Mining Software Repositories using Supervised Machine Learning technique.The work adopted Object Oriented Analysis and Design (OOAD) methodology for the system design and was implemented using PHP Hypertext Pre-processor scripting language for the purpose of enabling flexibility and userfriendliness in mining source codes from repositories.The database for the model was created using MySQL.The enhanced model utilized K-Nearest Neighbor (k-NN), a supervised machine learning algorithm for data classification to eliminate voluminous comments from source codes in order to reduce bulkiness.The results and performance evaluation of the existing and enhanced models were illustrated.The predefined parameters for both models comprised of the number of iterations for mining, the time taken to generate the codes in seconds and number of generated lines of codes.Seven iterations were carried out for both models in which the existing model generated a total of 56 lines of codes in 2.322 seconds, while the developed model generated a total of 93 lines of welldefined lines of codes in 0.017 seconds.Therefore, the results obtained clearly showed that the model performed much better than the existing model in terms of speed, accuracy and extraction of well defined codes.The model could be beneficial to data miners, programmers, software engineers, project managers of large industrial environments as well as researchers because relevant information from the study can be applied to problem-solving.