TY  - BOOK
AU  - Anupa K Narayanan (41718003)
TI  - Analysis of changing face of engineering education in Kerala 
KW  - EDUCATIONAL DATA MINING
KW  - EDUCATION
N1  - Education is no longer a onetime event but a lifelong experience.
The future of humanity depends very much on the cultural, scientic,
and technological developments which evolve from the centres
of Higher Education. The mission of a higher educational institution
( a college or a university) hence, is to be a pioneer in the creation
of an era so as to preserve the God given nobility of human existence,
recognizing its moral and spiritual dimensions. In a country
like India, along with the enormous growth in population size, more
and more higher educational institutions are established and hence
become more competitive in nature. These institutions are directed
towards providing quality education to its students.
In the current scenario a bench mark of quality in education is
achieved by analysing students performance using standard data
mining techniques. The advent of information technology has led to
large volumes of data storage in various formats like records, les,
documents, images, sound, videos, scientic data and many new
data formats. The data collected from dierent applications require 
proper method of extracting knowledge from large repositories for
better decision making. This is an extreme challenge for institutions
that uses traditional data management mechanism to store
and process huge datasets.
Reliable early stage predictions of a students future performance
could be critical to facilitate timely interventions during the course
and to prepare strategies in teaching and learning methods. Though
several statistical researches has been carried out extensively using
educational data, the availability of enormous computing powers,
and powerful algorithms have proven to be more eective in this
eld of data mining than the educational statistics problems. In the
current scenario, it becomes apparent that some forms of intelligent
system will be eective in assisting these advisors.
This work which we present here, details the process of collecting
organizing and mining student data from the institutes of higher
education, the Engineering Colleges across Kerala, aliated to the
Dr A P J Abdul Kalam Technological University or Kerala Technical
University(KTU).
The work investigates the in
uence of socio economic and non
academic factors that aect the performance of a student and presents
the frame work of an Student performance prediction system which
is a great area of concern for educational institutions to prevent their
students from failure followed by attrition, leading to the very poor
performance of several Engineering Colleges under KTU, by pro-viding necessary support and counselling to complete their degree
successfully.
In this work we more deeply investigate the direct utility of using
clustering to improve prediction accuracy and provide explanations
for why this may be so. We look at two datasets, run k-means
at dierent scales and for each scale predictors are trained. This
produces k sets of predictions. These predictions are then combined
by a nave ensemble. It was observed that this use of a predictor
in conjunction with clustering improved the prediction accuracy in
most datasets. This is believed to indicate the predictive utility of
exploiting structure in the data and the data compression handed
over by clustering. It was also found that this method improves upon
the prediction of even a Random Forests predictor, which suggests
this method is providing a novel, and useful source of variance in
the prediction process.
The scope of this research is to examine the accuracy of the ensemble
techniques for predicting the students academic performance,
particularly for four year engineering graduate program. The non
parametric ensemble methods called Random forests, Bagging and
AdaBoost are a powerful class of supervised algorithms for both classication
and regression. The model is mainly focussed on nding
the prediction accuracy of academic performance of students using
these methods. The model also bring to light the various student
attributes that are highly in
uential on the performance of a learner
ER  -