Aproximación metodológica para la Integración de DataOps en proyectos de analítica
Archivos
Autores
Rivera Olarte, David Fernando
Director
Tipo de contenido
Trabajo de grado - Maestría
Idioma del documento
EspañolFecha de publicación
2019-11-05
Título de la revista
ISSN de la revista
Título del volumen
Resumen
La aplicación de metodologías ágiles en conjunto con las prácticas continuas de DevOps son usadas en proyectos de analítica buscando presentar los resultados de los modelos de analítica de una manera más rápida, con más control del proceso de desarrollo y con mayor calidad. Esta es una tendencia emergente que ha sido llamada DataOps. Para que esta tendencia sea válida como proceso de entrega de datos, se hace necesario establecer un marco de actuación que siga los procedimientos tradicionales como CRISP-DM, e integre las herramientas de las prácticas continuas de DevOps y permita tener los beneficios de velocidad y calidad. La presente investigación presenta una propuesta metodológica donde se unifican las aproximaciones de la ingeniería de software y la analítica, entregando los resultados de una manera integrada y automatizada en su despliegue y monitoreo. Como principal contribución de la metodología aplicada se encuentra que una vez definidas las herramientas de integración, es posible contar con una doble orquestación, una para el entregable software y otro al producto de datos, donde se pueden controlar sus variables de calidad, velocidad de entrega y valor esperado del negocio.
Abstract: The application of agile methodologies together with the continuous practices of DevOps are used in analytical projects seeking to present the results of the analytical models in a faster way, with more control of the development process and with greater quality. This is an emerging trend that has been called DataOps. In order for this trend to be valid as a data delivery process, it is necessary to establish an action framework that follows traditional procedures such as CRISP-DM, and integrates the tools of continuous DevOps practices, allowing for speed and quality benefits. This research presents a methodological proposal where the approaches of software engineering and analytics are unified, delivering the results in an integrated and automated way in their deployment and monitoring. The main contribution of the applied methodology is that once the integration tools have been defined, it is possible to have a double orchestration, one for the deliverable software and another for the data product, where it is possible to control its quality, delivery speed and value Expected from the business.
Abstract: The application of agile methodologies together with the continuous practices of DevOps are used in analytical projects seeking to present the results of the analytical models in a faster way, with more control of the development process and with greater quality. This is an emerging trend that has been called DataOps. In order for this trend to be valid as a data delivery process, it is necessary to establish an action framework that follows traditional procedures such as CRISP-DM, and integrates the tools of continuous DevOps practices, allowing for speed and quality benefits. This research presents a methodological proposal where the approaches of software engineering and analytics are unified, delivering the results in an integrated and automated way in their deployment and monitoring. The main contribution of the applied methodology is that once the integration tools have been defined, it is possible to have a double orchestration, one for the deliverable software and another for the data product, where it is possible to control its quality, delivery speed and value Expected from the business.