An Efficient Data Mining Technique for Assessing Satisfaction Level with Online Learning for Higher Education Students

Main Article Content

M. Pragathi, Sharanya Kaleri, M. Sony, Saleha Sulthana

Abstract

By working on understudies' scholarly execution, each instructive association endeavors to raise the general nature of schooling. For this reason Educational Data Mining (EDM) is a quickly developing field of exploration that utilizes the standards of Data Mining (DM) to assist instructive foundations with finding significant data about Student Satisfaction Level (SSL) with regards to Online Learning (Old). To get the ideal informational settings, a couple of practices have been taken a stab at using EDM to calculate students' approaches to acting. Thus, Feature Selection (FS) is regularly used to distinguish the least cardinality yet most pertinent subset of attributes. This study examines the similarity of the SSL model with FS approaches inside and out in light of the fact that the FS cycle fundamentally affects a fulfillment model's anticipated precision. In such manner, a web-based poll was utilized to at first gather a dataset of understudy assessments of Old courses. The display of covering FS approaches in DM and portrayal computations was surveyed to the extent that health values using this dataset. At long last, the nature of 11 covering based FS calculations and the k-NN and SVM as gauge classifiers is utilized to assess the decency of subsets with different cardinalities concerning expectation precision and the quantity of chosen highlights. Both the best strategy and the element subset with the best dimensionality were tracked down in the examinations. The ongoing review's discoveries plainly support the deeply grounded association between an expansion in expectation exactness and the presence of few qualities. The fundamental assortment of attributes may really help with the improvement of productive educational drives, making FS very important for high-accuracy SSL expectation. Our work yields a component size decrease of up to 80% and order accuracy of up to 100%  on the used continuous dataset.

Article Details

Section
Articles