| dc.creator |
Mdenye, Rose Manfred |
|
| dc.date |
2019-08-18T09:25:50Z |
|
| dc.date |
2019-08-18T09:25:50Z |
|
| dc.date |
2015 |
|
| dc.date.accessioned |
2022-10-20T13:46:54Z |
|
| dc.date.available |
2022-10-20T13:46:54Z |
|
| dc.identifier |
Mdenye, R. M. (2015). Development of support vector machine model for prediction of students’ dropout in Higher Learning Institutions: A case of the University of Dodoma. Dodoma: The University of Dodoma. |
|
| dc.identifier |
http://hdl.handle.net/20.500.12661/769 |
|
| dc.identifier.uri |
http://hdl.handle.net/20.500.12661/769 |
|
| dc.description |
Dissertation (MSc Computer Science) |
|
| dc.description |
Students’ dropout from completing courses in higher learning institutions is a prevailing problem in Tanzania as in other countries. The intensity of the problem varies from one institution to another. Several measures have been taken by the institutions so as to reduce the number of students’ dropout. In this study, a support vector machine model is developed and tested in the prediction of the students’ dropout to help reduce the number of students dropping out of studies by predicting the risky students before they dropout. This study focused on the first year students who have a higher risk of the university dropout.
In developing the model, opinion documents from students were collected in which the first year students gave out their opinions about their first year experience at the university. Also questionnaires were used to gather factors which contribute to students’ dropout. Based on the analyzed factors for dropout, the documents were labelled as risky or non-risky. The labelled documents were grouped into two groups which are the training and test documents. The model was developed using the training documents and was implemented using algorithms written in C++ and the SVM light package. Testing of the model was done using the test documents. Performance evaluation of the model was done to check the accuracy of the model in prediction of students’ dropout.
The kernels used were the linear and polynomial kernel with different values of p and the radial basis function kernel with different values of gamma. The Radial Basis Function Kernel model outperformed the rest by recording an accuracy of 98.5% and the F score performance measure of 97.56%. |
|
| dc.publisher |
The University of Dodoma |
|
| dc.subject |
Higher learning institutions |
|
| dc.subject |
Student dropout |
|
| dc.subject |
Vector machine model |
|
| dc.subject |
Dropout |
|
| dc.subject |
Vector machine |
|
| dc.subject |
Learning institutions |
|
| dc.subject |
Dropout prediction machine |
|
| dc.title |
Development of support vector machine model for prediction of students’ dropout in Higher Learning Institutions: a case of the University of Dodoma |
|
| dc.type |
Dissertation |
|