Improvement of software engineering studies projects with data mining

The effects of the country?s economy and the growing importance of software in our lives have increased the interest of software engineering at both of the academic and industrial platforms. The Development of software applications is extremely expensive, error prone and requires hard work. Cause of failures in software projects; the planned budget or deadline exceeded the customer demands can not be met, the cancellation of projects, low confidence and to be out of service due to quality issues. The industry claims that the software engineering graduates are not able to meet their requirements due to the failures of software projects. This also creates a gap between academia and software industry. Aim of this study to improve software engineering course projects. Therefore we have applied questionnaire totally 189 students about courses projects, 86 students attended from Yıldız Technical University in System Analysis and Design course, 70 students attended from Yıldız Technical University in Software Engineering course and 33 students attended from Namık Kemal University in Agile Software Development course. According to the findings, students are not enough social skills such as documentation, team communication, and project time difficulties in the projects. Also the most successful algorithm is determined which is one of the methods often used in data mining using the classification for classification survey question. Accuracy, kappa, mean absolute error, root mean squared error results are taken into consideration for determining the most successful classification algorithm. According to the results, J48 algorithm is the most successful algorithm for classification the survey questions. In addition, find unexpected relationships between the survey questions, to reveal confidential information, apriori algorithm of association rule mining are used, which is one of data mining algorithms.