Predicting Whether Student will continue to Attend College or not using Deep Learning

Download Edit this record How to cite View on PhilPapers
Abstract
According to the literature review, there is much room for improvement of college student retention. The aim of this research is to evaluate the possibility of using deep and machine learning algorithms to predict whether students continue to attend college or will stop attending college. In this research a feature assessment is done on the dataset available from Kaggle depository. The performance of 20 learning supervised machine learning algorithms and one deep learning algorithm is evaluated. The algorithms are trained using 11 features from 1000 records of previous student registrations that have been enrolled in college. The best performing classifier after tuning the parameters was NuSVC. It achieved Accuracy (91.00%), Precision (91.00%), Recall (91.00%), F1-score (91.00%), and time required for training and testing (0.04 second). Additionally, the proposed DL algorithm scored: Accuracy (93.00%), Precision (93.00%), Recall (93.00%), F1-score (93.00%), time required for training and testing (0.66 second) for predicting whether student will continue to attend college or not.
Keywords
No keywords specified (fix it)
Categories
(categorize this paper)
PhilPapers/Archive ID
ABUPWS
Upload history
Archival date: 2022-07-01
View other versions
Added to PP
2022-07-01

Downloads
78 (#58,582)

6 months
78 (#9,150)

Historical graph of downloads since first upload
This graph includes both downloads from PhilArchive and clicks on external links on PhilPapers.
How can I increase my downloads?