Nowadays, the companies that develop software are more aware about the necessity to invest in testing and quality assurance. This is due the improvement that testing sector has done the last few years, investing in innovation and developing software more efficient and more customized. All the analytic testing data software in the market are focus in show a static analysis without any logic and process, this means that this software only generate an automatic graphic report with all data of the projects and his statistics. The software developed in this thesis, use a logical data collection through a customized filter for each of the projects, then process own standard model with the IBM SPSS server using machine learning. This server will make a predictions of an established parameters that depends of their results, will be change the project in planning phase reducing the initial project costs. In recent years, it has appeared the necessity of produce a software with features mentioned earlier, considering that actually there isn't any software with this specific features or similar. Therefore, there is a market opportunity which must be harnessed. The methodology used to develop this project has been Scrum. The dynamic of this methodology is the next one: first define the objectives of the project and classify the tasks and group in a list, named Release Backlog. During development will be make a two weeks of slots of time, the first day of the slot there is a meeting about the tasks that will be make in this two weeks (Sprint Planning), and finally the last day of the slot will be make another meeting to check the work (Sprint Review).