Sistema basado en instancias para pronosticar el éxito de un estudiante a su ingreso en una carrera universitaria
Fecha
2008-07-10
Autores
King, Anthony Omari
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Central “Marta Abreu” de Las Villas
Resumen
Los análisis acerca de los problemas de la retención en la Universidad Cubana apuntan la necesidad de fortalecer la atención individualizada de los estudiantes, y en especial en los primeros años de su vida en la universidad. Con esta finalidad el diagnóstico individual es un importante instrumento para conocer y a la vez trabajar de forma integral en la formación de cada educando, y en tales tareas ha tenido mucho éxito la aplicación de técnicas de la Inteligencia Artificial. Por otro lado, el proceso de descubrimiento de conocimiento en bases de datos está siendo cada vez más utilizado por las organizaciones como base para la toma de decisiones. Si en la etapa de “minería de datos” se aplican algoritmos de aprendizaje automático para definir modelos basados en técnicas de la Inteligencia Artificial, se obtendrán modelos comprensibles para los usuarios. En particular, la aplicación de modelos híbridos utilizando técnicas de Soft Computing han mostrado su efectividad porque al mismo tiempo que su aplicación elimina el cuello de botella que constituye la ingeniería del conocimiento en los sistemas inteligentes, el uso de la lógica borrosa provee una semántica para representar, manipular y utilizar la imprecisión presente en aplicaciones del mundo real.
Como parte de la informatización de la Universidad Cubana, en éstas se cuenta con un “Sistema para el Control de Estudiantes”. En este trabajo se muestran resultados del proceso de descubrir conocimiento a partir de la Base de Datos de este sistema, lo cual aporta información útil a la toma de decisiones en la gestión del proceso docente-educativo. En particular, se presenta una primera versión del producto de software “Sistema basado en instancias para pronosticar el éxito de un estudiante universitario” (SIPEXIS) que muestra resultados satisfactorios en la tarea de determinar el éxito de un estudiante de primer año a su ingreso en la Universidad. En la minería de datos fueron utilizados varios algoritmos implementados en la herramienta de aprendizaje automático Weka, posterior a una etapa de preparación de los datos asistida por la misma herramienta. Esta aplicación implementa el modelo ConFuCiuS, un modelo híbrido desarrollado en el grupo de Inteligencia Artificial de esta universidad, que muestra resultados satisfactorios para casos de control representativos de cinco de las especialidades que se estudian en la UCLV.
The analyses of retention problems in Cuban Universities point to the necessity to strengthen the individual attention given to students, especially in the first years of their university life. With this purpose the individual diagnosis is an important instrument to know and, at the same time, apply in an integral way in the formation of each student. In such tasks the application of Artificial Intelligence techniques has been very successful. On the other hand, the process of knowledge discovery in databases is being more and more used by organizations as a base for decision making. If in the “data mining” stage machine learning algorithms are applied to define models based on Artificial Intelligence techniques, understandable user models will be obtained. In particular, the application of hybrid models using Soft Computing techniques has shown its effectiveness because at the same time that the application eliminates the bottle neck that knowledge engineering constitutes in intelligent systems, the use of the fuzzy logic provides the semantics to represent, manipulate and use the imprecision present in real world applications. As part of the process of incorporating informatics in Cuban Universities, there is a "System for the Control of Students" in existence. In this thesis, the results of the knowledge discovery process of this system’s database are shown. This contributes useful information to decision making in the administration of the educational process. In particular, a first version of the software product "Instance based system to predict the success of a university student" (SIPEXIS – acronym of its Spanish name) is presented. It shows satisfactory results in the task of determining the success of a first year student when admitted into the university. In the data mining process, several algorithms implemented in the machine learning tool Weka were used after a data preparation stage assisted by the same tool. This application implements the ConFuCiuS model, a hybrid model developed in the Artificial Intelligence group of this university that shows satisfactory results for representative control cases of five of the specialties studied at UCLV.
The analyses of retention problems in Cuban Universities point to the necessity to strengthen the individual attention given to students, especially in the first years of their university life. With this purpose the individual diagnosis is an important instrument to know and, at the same time, apply in an integral way in the formation of each student. In such tasks the application of Artificial Intelligence techniques has been very successful. On the other hand, the process of knowledge discovery in databases is being more and more used by organizations as a base for decision making. If in the “data mining” stage machine learning algorithms are applied to define models based on Artificial Intelligence techniques, understandable user models will be obtained. In particular, the application of hybrid models using Soft Computing techniques has shown its effectiveness because at the same time that the application eliminates the bottle neck that knowledge engineering constitutes in intelligent systems, the use of the fuzzy logic provides the semantics to represent, manipulate and use the imprecision present in real world applications. As part of the process of incorporating informatics in Cuban Universities, there is a "System for the Control of Students" in existence. In this thesis, the results of the knowledge discovery process of this system’s database are shown. This contributes useful information to decision making in the administration of the educational process. In particular, a first version of the software product "Instance based system to predict the success of a university student" (SIPEXIS – acronym of its Spanish name) is presented. It shows satisfactory results in the task of determining the success of a first year student when admitted into the university. In the data mining process, several algorithms implemented in the machine learning tool Weka were used after a data preparation stage assisted by the same tool. This application implements the ConFuCiuS model, a hybrid model developed in the Artificial Intelligence group of this university that shows satisfactory results for representative control cases of five of the specialties studied at UCLV.
Descripción
Palabras clave
Sistema Basado en Instancias, Pronostico de Éxito, Estudiantes, Ingreso en la Universidad, Bases de Datos, Weka, Inteligencia Artificial