Software engineering is an important concept for the development of information systems, which is more than coding or writing of programs, that it includes quality, scheduling, economics and knowledge of the application of principles. Software is subject to errors in manufacturing and operation and needs to be evaluated like any other product. In this research paper, we present a development of the traditional programmatic model of software engineering using the CRISP-DM methodology, which is primarily used in data extraction engineering systems, that aims to improve software quality. We study the effect of introducing this methodology in terms of reducing the number of errors discovered in software and its impact on error detection at an early stage of the software life cycle. The results showed that the use of this methodology helped to reduce the errors discovered in the software after the development process. It also helped to detect a greater percentage of errors in the early stages of the software life cycle, which helps to increase the quality of the software, compared to a relatively low number of lines. Code because of applying the proposed methodology.