Prediction Model for Students’ Future Development by Deep Learning and
Tensorflow Artificial Intelligence Engine
Wilton W.T. Fok
1
, Y.S. He
1
, H.H. Au Yeung
1
, K.Y. Law
1
, KH Cheung
1
, YY. Ai
1
, P. Ho
1
1
The University of Hong Kong
Abstract—Classification and prediction of students’
performance in examination are the typical challenges for
educators. Various traditional data mining methods such as
decision tree and association rules were used to perform
classification. In recent years, the rapid development of
artificial intelligence and deep learning algorithm provided
another approach for intelligent classification and result
prediction. In this paper, a research on how to use Tensorflow
artificial intelligence engine for classifying students’
performance and forecasting their future universities degree
program is studied. An appropriate and accurate forecast is
important for providing prompt advice to student on program
and university selection. For a more comprehensive
consideration of an all rounded factors, the deep learning
model analysed not only the traditional academic performance
including Mathematic, Chinese, English, Physics, Chemistry,
Biology and History, but also non-academic performance such
as service, Conduct, Sport and Art. A few parameters in
Tensorflow engine including the number of intermediate nodes
and number of deep learning layers are adjusted and
compared. With a data set of two thousands students, 75% of
these data are used as the training data and 25% are used as
the testing data, the accuracy ranged from 80% to 91%. The
optimal configuration of the Tensorflow deep learning model
that achieves highest prediction accuracy is determined. This
study determined the factors affecting the accuracy of the
prediction model.
Keywords: e-Learning assessment, Artificial Intelligence,
Deep Learning, prediction modelling
I. INTRODUCTION
How to predict students’ performance is always a
question concerned by the students’ teachers and parents.
Based on the past examination results and in-class
assessments, it is possible to forecast the future development
of the students. It is a challenging and important matters as it
involves the large volume of data in educational databases
and the result could impact the future development of a
young kid. A good and accuracy prediction could bring the
benefits and impacts to students, educators and academic
institutions. Various type of data mining techniques had
been used for performance prediction for decades, e.g.
decision tree, Naive Bayes, K-Nearest Neighbor and Support
Vector Machine [1]. However, with the rise of artificial
intelligence and deep learning application, using AI engine
such as Google Tensorflow for pattern recognition has now
been rising it importance. In this paper, we will investigate
how to use artificial intelligence and deep learning algorithm
for pattern recognition and correlation of assessment results.
There are some traditional data mining techniques that have
been used to predict students’ performance. Some researches
educational data mining method had been done to identify
those important attributes in students data.
II. T
RADITIONAL METHOD FOR CLASSIFICATION AND
PREDICTION
In order to build the predictive modeling, there are
several traditional tasks used for example classification,
regression and categorization. Algorithm such as association
rules and decision tree are commonly used.
Association rule reflects the interdependence and
correlation between different things. It is commonly used in
physical stores and e-commerce recommendation systems.
Likewise, it can be used for recommending schools and
subjects to students. Support, confidence are two key
concepts of them. In a given database, each transaction
contains a set of items. Every rule is composed by two
different itemsets X and Y. Support is the percentage of the
transaction contains both X and Y, confidence is the
percentage of Y that contains X. In other words, support is
the probability while confidence is the conditional
probability. If the minimum thresholds of them, which have
to be set by us, are satisfied, there is an association that
exists between X and Y. Association rule mining is more
applicable to the situation where the index in the record are
discrete value. If the indicator values in the original database
are continuous data, and appropriate discretization should be
performed previously.
The classification decision tree model is a tree structure
that describes the classification of instances. The decision
tree consists of nodes and directed edges. There are two
types of nodes: internal nodes and leaf nodes, internal nodes
represent a feature or attribute, and leaf nodes represent a
class.
When categorizing, a certain feature of the instance is
tested starting from the root node, and the instance is
assigned to its child node according to the test result; at this
time, each child node corresponds to a value of the feature.
This recursively moves down until it reaches the leaf node,
and finally assigns the instance to the class of the leaf node.
The decision tree can be seen as a set of if-then rules: a rule
is constructed from the root node of the decision tree to each
path of the leaf nodes; the characteristics of the internal
nodes on the path correspond to the conditions of the rules,
and the leaf nodes Corresponds to the classification of the
conclusions.
2018 4th IEEE International Conference on Information Management
e-mail: wtfok@eee.hku.hk
103
978-1-5386-6147-5/18/$31.00 ©2018 IEEE