Post on 27-Aug-2014
description
EDUCATIONAL DATA MINING
In Relation To Educational Statistics Of Nepal
Final Presentation
Presenters:Roshan Bhandari (16226)Sijan Bhandari (16236)Subit Raj Pokharel (16237)Sujit Maharjan (16239)
Supervisor:Bibha Sthapit(Lecturer, Pulchowk Campus)
Co-Supervisor:Anjesh Tuladhar(COO, YIPL)
35 000 Schools50 billion + budget
20 parameters2 times a year
400 + NGOs and INGOs
Statistics
Department of Education
Office of Controller of Examination
“Education for all”
Problem ???
Reports Not Accessible
Unmanaged and Unformatted
2007Schools - 29448Stds - 6533392
2008Schools - 31156Stds - 6964553
2009Schools - 32130Stds - 7295433
2010Schools - 33160Stds - 7463793
2011Schools - 34361Stds - 7444134
Educational Data Mining
• New methodology of manipulating educational data.
• Modern methodology started since 2008.
• In this regard, we are the first people to introduce Educational Data Mining to Nepal
What we have done ??
Block Diagram
API For Developers
District wise number of school in the year 2007
http://edm.hamroschool.org/numschool/2007
API For Developers
Parameter values of Kaski for 2011
http://edm.hamroschool.org/kaski/2011
Map
Charts
Districts with max. Dropout Rate
Educational Development Index (EDI)
Map
Educational Development Index for Individual
District
Comparison of DistrictsYear wise Pupil Teacher Ratio Comparison
Cluster Analysis
Year2066 Year2067 Year20681 24.08481 26.19629 25.746092 53.66839 50.65694 52.70283
Government 2785
private 87
Cluster Analysis
Government 2872
Prediction & Parameter Relationships● Correlation Coefficient● Least Square Regression● Standard Error of Estimate● Multiple Regression
Multiple Regression
Our proposed models:Gross Enrollment ModelGirls Enrollment ModelDropout Rate Model:
○ Dropout rate = 53.65 + 0.4644 * Enrolled Passed - 5.455*GPI - 0.084*Unqualified Teacher
○ Standard Error = 2.87
○ R2 = 80.73 %
Schools Classification• Tried to classify schools based on the
performance of the schools
• Six subjects, average marks of school and pass out data are used to classify schools
• ID3 algorithm has been used to construct Decision Tree.
Decision Tree
Enhancement• More parameter can be included to get more
accurate results• More data can be provided by collaborating
with DoE• A real time data update and feeding
mechanism can be built to make system more realistic.
• More data mining techniques can be used to extract core information.
EDIClustering
RegressionClassification
Conclusion
Tools Used
Thank You