A Classification Model for Predicting Campus Placement
performance Class using Data Mining Technique.
K.Manikandan1, S.Sivakumar2, M.Ashokvel3
1
Assistant Professor, Department of Computer Applications,A.V.C. College of
Engineering,Mayiladuthurai, Tamilnadu, India.
2
Assistant Professor, Department of Computer Applications,A.V.C. College of
Engineering,Mayiladuthurai, Tamilnadu, India.
3
Assistant Professor, Department of Computer Applications,A.V.C. College of
Engineering,Mayiladuthurai, Tamilnadu, India.
ABSTRACT
Data Mining is very useful in the field of education to predict campus placement performance of the students.
Placement of students is one of the important activities in educational institutions. Admission and reputation of
institutions mainly depends on placements. The overall goal of data mining is to extract information from a
dataset and transform it into useful structure for further use. This can help in building new systems and take
decision making in educational system. This paper discusses use of classification algorithms in educational data
mining. Classification algorithms is applied on student previous year data to compute the distribution of
placement class and it can be used to predict the students' placement in various companies. This will help the
students to identify the category of company in which they are eligible and prepare accordingly in an efficient
manner.
Keywords —Data Mining, Classification, Naive Bayesian, placement Prediction
I .INTRODUCTION
Data mining is the process of discovering interesting knowledge from large amount of data stored in database,
database warehouse or other information responsibility. Data mining term is mainly used for the specific of six
activities namely Classification, Estimation, Prediction, Association rules, Clustering, Description and
Visualization. The first three tasks – classification, estimation and prediction are all examples of directed data
mining or supervised learning. Majority of students in higher education join a course for securing a good job.
Therefore taking a wise career decision regarding the placement after completing a particular course is crucial in
a student’s life. Data mining is one of the important techniques used in Education field. In real world, predicting
the performance of the student’s placement is a challenging task. The primary goals of Data Mining in practice
tend to be Prediction and Description. Predicting performance involves variables like X Mark, XII Mark, UG
Mark, PG Mark Programming language, etc. in the student database to predict the unknown or future values of
interest. Educational Data Mining uses many techniques such as Decision Trees, Multilayer Perception,
NaïveBayes and many others. Using these methods many kinds of knowledge can be discovered. The aim of
classification is to predict the future output based on the available data. The prediction of computer science
students where they can be placed after the completion of their course will help to improve efforts of students
for proper progress. It will also help teachers to take proper attention towards the progress of the student during
29 | P a g e
the course. It will help to build reputation of institute in existing similar category institutes in the field of IT
education. The present study concentrates on the prediction of placements of Computer Science students. We
apply data mining techniques using J48 Algorithm, REPTREE Algorithm, Naïve Bayes classifier,BayesNet
classifier, Multilayer Perceptron to interpret potential and useful knowledge.
II EDUCATIONAL DATA MINING
Educational data mining is an emerging discipline, concerned with developing methods for exploring the
unique types of data that come from educational settings and using those methods to better understand
students performance. Different from data mining methods, EDM, when used explicitly, accounts for (and avail
of opportunities to exploit) the multilevel hierarchy and lacks independent educational data.Educational data
mining methods come from different literature sources including data mining, machine learning, psychometrics,
and other areas of computational modelling, statistics, and information visualization.
III CLASSIFICATION ALGORITHMS
Educational Data mining can be implemented in many techniques such as decision trees, neural networks, k-
nearest Neighbor, Naïve Bayes, support vector machines and many others. Using these methods many kind of
knowledge can be discovered such as association rules, classification, clustering, and pruning the data. The
classifiers used in this paper consists of common decision tree algorithm C4.5 (J48), REPTREE Algorithm,
Naive Bayes classifiers, BayesNet classifiers and Multilayer Perceptron (MLP) algorithm. The results obtained
from the classification task are presented in the experimental results.
A. J48 Algorithm
The J48 classification algorithm is WEKA’s version of the implementation of the C4.5 decision tree algorithm,
which uses a greedy technique to induce decision trees and make use of reduced- error pruning. The algorithm
was developed from ID3 algorithm for handling missing data, continuous data, pruning, splitting and generating
rules [12]. The technique uses Gain Ratio instead of Information Gain for splitting purpose:
In order to categorize a given set, Information Gain as a metric is compulsory, with a function to deliver a
balance inthe splitting. Providing a data set that containsattributes, we can measure the entropy as a degree
ofimpurity
30 | P a g e
And determining the best attribute for a node in the tree, we use the Information Gain as a measure, such that
Information Gain, Gain (S,A) attributes are defines as:
B. REPTREE Algorithm
The REPTREE classification algorithm is a technique that builds trees using entropy as impurity measure and
also makes use of reduced-error pruning.
C. NAIVE BAYES Algorithm
The algorithm is based on Bayes rule of provisional possibility and adopts independence between attributes
values in a data set [14]. The algorithm requires small amount of training data to predict a classification model.
The technique signifies a method to probabilistic discovery of knowledge and gives efficient algorithm for data
classification [15]. The algorithm makes use of the Bayesian theorem with naïve independent assumptions as in
the formula [3]
D. BayesNet Classifiers
A Bayesian network, Bayes network model or probabilistic directed acyclic graphical model is a probabilistic
graphical model (a type of statistical model) that represents a set of variables and their conditional dependencies
via a directed acyclic graph (DAG).
E. Multilayer Perceptron
Multilayer Perceptron (MLP) algorithm is one of the most widely used and popular neural networks. The
network consists of a set of sensory elements that make up the input layer, one or more hidden layers of
processing elements, and the output layer of the processing elements. MLP is especially suitable for
approximating a classification function (when we are not so much familiar with the relationship between input
and output attributes) which sets the example determined by the vector attribute values into one or more classes.
31 | P a g e
IV DATA COLLECTION
Table 1: The definition of attributes and values
Values
Attribute Definition
Roll number Identity of a student ROLL NO
SSLC X Standard board marks 1 – 100
HSC Higher secondary board marks 1 – 100
UG Under graduate marks 1 – 100
PG Post graduate marks 1 – 10
Specialization Programming Knowledge Java, .Net
Company Level Company Details about
T1 , T2, T3
selection in campus interview
In this case, we have collected 2013-2016 batch students’ details as a dataset from computer application
department in A.V.C. College of Engineering. Here, Company Status T1 refers to Tier1 level company, T2
refers to Tier2 level company and T3 refers to Tier3 level company.
V DATA PREPROCESSING
Raw data is a quality less and inconvenience data for processing. This poor quality of raw data affects the data
mining efficiency. In order to improve the quality of the data and, also the mining results pre-processing of raw
data is carried out. Data preparation and filtering steps takes large amount of time. In this case, we have
collected computer application students details as a dataset. Here, irrelevant attributes such as students
residential address, name, etc had been removed.
VI IMPLEMENTATION OF MINING MODEL
Weka is open source software that implements a large collection of machine leaning algorithms and is widely
used in data mining applications. From the above data, placement.arff file was created. This file was loaded into
WEKA explorer. The classify panel enables the user to apply classification and regression algorithms to the
resulting dataset, to estimate the accuracy of the resulting predictive model, and to visualize erroneous
predictions, or the model itself. The algorithm used for classification is J48 Algorithm, Naive Bayes, BayesNet,
REPTree and Multilayer Perceptron (MLP). Under the "Test options", the 10-fold cross-validation and
percentage split are selected as our evaluation approach. Since there is no separate evaluation data set, this is
necessary to get a reasonable idea of accuracy of the generated model. This predictive model provides way to
predict whether a new student will place or not in an organization.
32 | P a g e
VII EXPERIMENTAL RESULTS AND DISCUSSIONS
The objective of the study is to explore if it is possible to predict the student placement based on the various
input variables which are retained in the model. The classification model was built using several different
algorithms and each of them using different classification techniques. The WEKA Explorer application is used
at this stage. Each classifier is applied for two testing options – cross validation and percentage split. The screen
shot of the WEKA preprocessing stage is shown in Figure 1.
Figure 1. WEKA Screenshot of Data Distribution in the Preprocessing Stage.
Table 2. Classification results for the decision tree algorithm
J48 – 10-fold Cross validation J48 – Percentage split
Class TP Rate Precision TP Rate Precision
T1 0.889 0.857 0.909 1
T2 0.85 0.68 0.75 0.375
T3 0.652 0.882 0.556 0.833
Weighted Avg. 0.8 0.815 0.75 0.833
7.1 Results of Decision Tree Classifier
In this study, J48 classification algorithm was implemented on the data and the results of the classification is
presented in Table 2. It is inferred from the Table 2, that J48 has correctly classified about 80% for the 10-fold
cross-validation testing and 75%for the percentage split testing. It produces a classification tree. The screenshot
of decision screen building process in shown in Figure 2. The results from Table 2 reveal that the True Positive
Rate is high for the classes – T1 (90 %) and T2 (75-85 %).
33 | P a g e
Figure 2. Screenshot of Decision Tree build using J48 Classifier.
Table 3. Classification results for the REPTreeAlgorithm
REPTree– 10-fold Cross validation REPTree– Percentage split
Class TP Rate Precision TP Rate Precision
T1 0.852 0.92 0.73 0.75
T2 0.65 0.52 0.75 0.188
T3 0.652 0.75 0.444 1
Weighted Avg. 0.729 0.75 0.417 0.75
7.2 Results of REPTreeClassifier
Table 3 shows the classification results for REPTreeclassifier. The REPTreeclassifier correctlyclassifies about
72.9% for the 10-fold cross-validation testing and 41.7 % for thepercentage split testing. The results from Table
3 show that the True PositiveRate is high for the T1 (73-85%) and T2(65-75%).The precision is found to be
high for theclasses – T1 and T3.
Table 4. Classification results for the Naive Bayes Classifiers
Naive Bayes– 10-fold Cross Naive Bayes– Percentage split
validation
Class TP Rate Precision TP Rate Precision
T1 0.852 0.821 0.909 1
T2 0.65 0.722 0.75 0.6
T3 0.87 0.833 0.889 0.889
Weighted Avg. 0.8 0. 797 0.875 0.892
34 | P a g e
Table 5. Classification results for the BayesNet Classifiers
BayesNet– 10-fold Cross validation BayesNet– Percentage split
Class TP Rate Precision TP Rate Precision
T1 0.889 0.923 0.909 0.909
T2 0.7 0.667 0.75 0.429
T3 0.783 0.783 0.667 1
Weighted Avg. 0.8 0.804 0.792 0.863
7.3 Results of Bayesian Classifiers
The present study implements Bayesian classifiers namely Bayesian networks and naive Bayes on the dataset
and the results are presented in Table 4 and Table 5. Table 4presents the classification results for Naive Bayes
classifier and it is found that Naive Bayes classifier correctly classifies about 80 % for the 10-fold cross-
validation testing and 87.5% for the percentage split testing.The results from Table 3 reveal that the True
Positive is high for most of the classes – T3, T1 and T2. The precision is also high for the classes –T1, T3 and
T2. Table 5 presents results of BayesNet classifier on the dataset. It can be verified that Bayes Net correctly
classifies about 80 % for the 10-fold cross-validation testing and 79.2 % for the percentage split testing.The
results from Table 5, shows that the True Positive Rate is high for the classes –T1, T3 and T2. The precision is
also high for the classes – T1.
Table 6. Classification results for the MLP Classifiers
MLP – 10-fold Cross validation MLP – Percentage split
Class TP Rate Precision TP Rate Precision
T1 0.852 0.958 0.818 0.9
T2 0.65 0.619 0.75 0.5
T3 0.826 0.76 0.889 1
Weighted Avg. 0.786 0.796 0.833 0.871
7.4 Results of Multilayer Perceptron Classifier
The present study implements MLP on the dataset and the results are presented in Table 6. It presents the
classification results for MLP classifier correctly classifies about 78.6 % for the 10-fold cross-validation testing
and 83.3% for the percentage split testing. The True Positive is high for most of the classes – T1 and T3. The
precision is also high for the classes – T1 and T3.
35 | P a g e
1
0.9
0.8
0.7
J48
0.6
REPTree
0.5
Naive Bayes
0.4
BayesNet
0.3
0.2 MLP
0.1
0
Tier 1 Tier 2 Tier 3 Weighted Avg.
Figure 3. Classification algorithms performance comparison.
The results of the classification reveals that the Bayesian classifiers performs very well in comparison with other
classifiers with the highest overall accuracy, followed by J48 Classifiers. REPTree classifier and MLP Classifier
performs poorly and are less accurate than the others. The overall accuracy of all the tested classifiers is well
above 70%. Naïve Bayes and BayesNet registered accuracy for 80 %. J48 produces accuracy very near to 80 %.
In addition, further detailed analysis of the classification accuracy for the different classes reveals that the
predictions are worst for the Tier 2 class. The classification accuracy is very good for Tier 1 class.
VIII CONCLUSION
The J48, Naive Bayes, BayesNet, REPTree and Multilayer Perceptron (MLP)data mining techniques were
implemented on students data for analyzing the students placement selection..The results of the data mining
algorithms for the classification of the students based on the attributes selected reveals that the prediction rates
are not uniform among the algorithms. The range of prediction varies from 72-80 %. Bayesian classifiers
perform very well in comparison with other classifiers with the highest overall accuracy. Moreover, the
classifiers perform differently for the three classes. The data attributes that are found to have significantly
influenced the classification process are Tier 1 and Tier 3 classes. The study can be further extended to
performance of other classification techniques with larger sample dataset.
REFERENCES
[1] David Hand, HeikkiMannila, Padhraic Smyth Principles of Data Mining
[2] S. K. Mohamad and Z. Tasir, “Educational data mining: A review,” Procedia-Social and Behavioral
Sciences vol. 97, pp. 320–324, 2013
36 | P a g e
[3] C. Romero, S. Ventura, M. Pechenizkiy, and R. S. Baker, Handbook of educational data mining. CRC
Press, 2010
[4] Badr, E. Din, and I. S. Elaraby, “Data Mining : A prediction for Student ’ s Performance Using
Classification Method,” World J. Comput. Appl. Technol., vol. 2, no. 2, pp. 43–47, 2014.
[5] K. Bhardwaj, “Data Mining : A prediction for performance improvement using classification,” Int.J.
Comput. Sci. Inf. Secur., vol. 9, no. 4, 2011.
[6] E. Review, “Data mining approach for predicting student performance,” J. Econ. Bus., vol. X, no. 1, pp.
3–12, 2012.
[7] P. Ajith, B. Tejaswi, and M. S. S. Sai, “Rule Mining Framework for Students Performance Evaluation,”
Int. J. Soft Comput. Eng., vol. 2, no. 6, pp. 201–206, 2013.
[8] K. Daimi and R. Miller, “Analyzing Student Retention with Data Mining.,” in International Conference
on Data Mining, 2009, pp. 55–60.
[9] Baker RSJD. Data mining for education. International encyclopedia of education. 2010; 7:112–8.
[10] Pal AK, Pal S. Analysis and Mining of Educational Data for Predicting the performance of Students.
International Journal of Electronics Communication and Computer Engineering. 2013; . 4(5):1560–5.
[11] Rathee A, Mathur RP. Survey on Decision Tree Classification algorithm for the Evaluation of Student
Performance. International Journal of computers & Technology. 2013;4(2):244–7.
[12] Aher SB, Lobo LMRJ. Data mining in educational system using Weka. IJCA Proceedings on
International Conference on Emerging Technology Trends (ICETT). 2011; 3:20–5.
[13] Ajith P,,Tejaswi B, Sai MSS. Rule Mining Framework for Students Performance Evaluation.
International Journal of Soft Computing and Engineering. 2013; 2(6):201–6.
[14] Anuradha., C and T. Velmurugan, A Comparative Analysis on the Evaluationof Classification
Algorithms in thePrediction of Students Performance, Indian Journal of Science and Technology, Vol
8(15), July 2015
[15] Samrat Singh, Dr. Vikesh Kumar” Classification of Student’s data Using Data Mining Techniques for
Training & Placement Department in Technical Education”, International Journal of Computer Science
and Network (IJCSN), Volume 1, Issue 4, August 2012.
[16] Ajay Kumar Pal, Saurabh Pal “Classification Model of Prediction for Placement of Students”,
I.J.Modern Education and Computer Science, 2013, 11, 4956Published Online November 2013 in
MECS.
[17] NeelamNaik, SeemaPurohit“Prediction of Final Result and Placement of Students using Classification
Algorithm” International Journal of Computer Applications (0975 – 8887) Volume 56– No.12, October
2012
[18] V.Ramesh, P.Parkavi, P.Yasodha“ Performance Analysis of Data Mining Techniques or Placement
Chance Prediction” International Journal of Scientific & Engineering Research Volume 2, Issue 8,
August-2011 1 ISSN 2229
37 | P a g e
[19] A. K. Pal, and S. Pal, “Analysis and Mining of Educational Data for Predicting the Performance of
Students”, (IJECCE) International Journal of Electronics Communication and Computer Engineering,
Vol. 4, Issue 5, pp. 1560-1565, ISSN: 2278-4209, 2013.
[20] B.K. Bharadwaj and S. Pal., Data Mining: A prediction for performance improvement using
classification”, International Journal of Computer Science and Information Security (IJCSIS), Vol. 9,
No. 4, pp. 136-140, 2011.
38 | P a g e
