Thesis PERFILAMIENTO DE ESTUDIANTES Y ESCUELAS A TRAVÉS DE MODELOS DE APRENDIZAJE AUTOMÁTICO PARA LA ELECCIÓN DE ESCUELAS EN CHILE
Loading...
Date
2018-06
Authors
VARGAS POBLETE, CARLOS ANDRÉS
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Las políticas educacionales buscan asignar recursos públicos al desarrollo de establecimientos educacionales para lograr mejorar el aprendizaje de los niños. Dado que los recursos son escasos, se debe focalizar su asignación en los grupos más necesitados. Actualmente, los colegios se clasifican por su dependencia administrativa y los alumnos por su nivel socioeconómico. Esta separación unidimensional no representa adecuadamente la realidad. Los algoritmos de agrupamiento pueden ayudar a definir los programas educativos y sus beneficiarios. Se usaron algoritmos de X-Means modificados, logrando determinar cuatro clúster óptimos para cada entidad. Para los establecimientos los atributos más importantes son la dependencia y el nivel de copago que presentan. En el caso de los estudiantes los atributos más destacados dependerán del enfoque que se le dé, estos pueden ser por un lado el género y el nivel socioeconómico o el tipo de colegio al que asisten y el nivel de copago.
Educational policies seek to assign public resources to educational establishments for development, in order to improve children’s learning. Given that resources are limited, the designation should be focused on the most vulnerable groups. Currently, schools are classified by how they are supported financially and students by their socioeconomic level. This one- dimensional segregation does not represent reality accurately. Clustering algorithms may help define educational programs and their beneficiaries. Using X-Means modified algorithms; four optimal clusters were identified for each entity. For educational establishment the most relevant attributes were financial capacity of the school, private or public, and tuition expenses. For students attributes depend on the perspective, being the most relevant, gender and socioeconomic level or type of school and tuition.
Educational policies seek to assign public resources to educational establishments for development, in order to improve children’s learning. Given that resources are limited, the designation should be focused on the most vulnerable groups. Currently, schools are classified by how they are supported financially and students by their socioeconomic level. This one- dimensional segregation does not represent reality accurately. Clustering algorithms may help define educational programs and their beneficiaries. Using X-Means modified algorithms; four optimal clusters were identified for each entity. For educational establishment the most relevant attributes were financial capacity of the school, private or public, and tuition expenses. For students attributes depend on the perspective, being the most relevant, gender and socioeconomic level or type of school and tuition.
Description
Catalogado desde la version PDF de la tesis.
Keywords
ALGORITMOS DE AGRUPAMIENTO , ANALISIS DE DATOS , APRENDIZAJE NO SUPERVISADO , X-MEANS