Supervised Models for Measuring Performance At E-Learning Environment
C. S. Sasikumar1, A. Kumaravel2
1C. S. Sasikumar, Research Scholar in Bharath Institute of Higher education and Research, Chennai, India.
2A. Kumaravel, Professor and Dean, School of Computing, Bharath Institute of Higher education and Research, Chennai, India.
Manuscript received on 25 August 2019. | Revised Manuscript received on 07 September 2019. | Manuscript published on 30 September 2019. | PP: 2791-2797 | Volume-8 Issue-11, September 2019. | Retrieval Number: K22790981119/2019©BEIESP | DOI: 10.35940/ijitee.K2279.0981119
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: E-learning data becomes ‘Big’ data as it describes a huge volume of both structured and unstructured data. And inherent limitations of relational databases maintained in this context makes difficult to apply and to extract outputs meaningful. Data modeling is also recommended to design data views at various levels either conceptual or physical here. Most of the educational organizations are keen in collecting, storing and analyzing the students’ data because it will add more significant value to the decision making process. Data modeling through entity relationship model or query views plays a important role in dealing with big data due to the fact around 85% of big data is semi structured data. Hence data modeling should be carried out as required by any learning institution needs. Making big data component to reside in the data model is challenging. This paper is to establish data modeling techniques applied to a reasonably ‘big’ data in e-learning. Prediction models generated from this data will be accurate if the training sets and testing sets are governed properly in spite of data size complexity. Student Performance by study credits (partitioned in three classes: low, medium, high ) are classified with respect to their engagement attributes (activity types, sum of clicks made, duration in days) and obtained maximum accuracy 90.923%.
Keywords: E-learning, ER diagram, R Programming, SQL, Weka 3.9, Classifiers, J48, Jrip, Random Forest , Random Tree, Bagging. Big data.
Scope of the Article: Smart Learning Methods and Environments