Collaboration study case between the industry and the university: Use of lightgbm algorithm for data analytics in the execution history of scheduled routines (job)

At present, the link between the University and the industry for the generation of innovation is becoming more frequent. This link is achieved through cooperation projects, where a company presents a challenge to the university. In the case of computer engineering, the challenges are in the development of software, systems auditing or data analytics, among others. In this paper, we will present the work done by the university for a company. The objective of this project was to analyze a set of more than 5 million data to predict whether a Job (routine program to execute an executable) will be executed correctly or not. For the project, CRISP-DM was used as a methodology, and the activities carried out during the execution of the project range from the understanding of the business to the validation of the selected model. The algorithm presented for the proposed model was LightGBM, which has been widely used due to the speed of training with large amounts of data.