Data Mining Techniques To Predict Instructor Performance Khurram Abbas

Data Mining Techniques To Predict Instructor Performance

Khurram Abbas , Mr. Waqar Ahmad
Department of Computer Science
National Textile University
Faisalabad, Pakistan.
[email protected]

Best services for writing your paper according to Trustpilot

Premium Partner
From $18.00 per page
4,8 / 5
Writers Experience
Recommended Service
From $13.90 per page
4,6 / 5
Writers Experience
From $20.00 per page
4,5 / 5
Writers Experience
* All Partners were chosen among 50+ writing services by our Customer Satisfaction Team

Data mining techniques are very helpful tool used for understanding and solving the educational and administrative problems in educational field. Most of the research in this area is done on student’s performance. In this paper the main focus is that which data mining techniques are used to analyze instructor performance. We can use different classification techniques decision tree algorithms, Rule based algorithms, Naïve Bayes, C5.0 and CART. Furthermore we can also discuss about some techniques (Questionnaire method, generic web application based feedback model using unified modeling language and opinion mining) which are helpful to collect the student’s opinions about the teacher’s performance. The finding of the research is the importance and effectiveness of data mining techniques in educational field.
Data collection methods, Decision tree algorithms, Naïve Bayes, Educational data mining, measuring performance techniques, Opinion Mining
Now-a-days the most important challenge faces in educational domain is the proliferation of educational data and how we can use this data to improve the quality of education and different academic programs. How we can use this data for managerial decisions. In the existing educational system different kinds of formal and informal techniques are used for measuring the qualitative and quantitative work. These techniques are predefined quires and charts that do not find the useful hidden information. The best way to analyze useful hidden information is the use of data mining techniques.

Data mining is the process of discovering the hidden patterns and knowledge from a large amount of data set. Data mining can also be defines as the automated process of extracting useful information and knowledge including pattern, association, changes, trends, anomalies and significant structures that are unknown from large data set. Popularity of data mining techniques in educational domain is increases day by day because with the help of data mining techniques we can increase the student’s performance, increase the retention of student, decrease the dropout of students, take managerial decisions, academic planning, improve the performance of alumni department and also increase the instructor performance that is also our main focus.
One of the most important and common problem in education is to measure the instructor performance. This measurement is done through the Questionnaire method, generic feedback model using unified modeling language (UML) and opinion mining in a live course with the help of student’s evaluation about the course. The major problem in this measurement is that students did not evaluate the instructor performance correctly. But students are the only source of information in this technique and only student’s opinions are used for the effectiveness and the efficiency of the instructor.
The purpose of this research is to show the importance of data mining techniques in educational domain. In this study we discuss about Different techniques that can be used for analyzing the performance of instructor and also discuss that which technique is used in which type of data in which situation. This research may helpful for the peoples who are interested in that area of research (Educational Data Mining).
The rest of the paper is organized as follows: Section-II gives a review about related work in that domain. Section-III gives the information about the techniques which we can use to collect data from student’s feedback and also the techniques used for analyzing the instructor performance. Section-IV gives us the conclusion and future wok and Section-V is the references that we can use in this paper.
Literature review gives an innovative idea of the selected research topic to the researcher. It is the theoretical base of the research topic.
Lawrance et al. 8 present a research paper on the student’s performance using educational data mining a very important domain about data mining. In this research paper Lawrance use different types of attributes to know about the details of students. The attribute he can use in this research paper was student name, roll number, previous semester marks, assignments, attendance, gender, seminar performance and lab work. In this research paper different classification techniques are used and conclusion is that C5.0 gives good classification accuracy.
Agaoglu et al. 9 present different types of classification techniques SVM, DT, artificial neural network are used to build a classifier model. The performance of this classifier model is calculated on the basis of different performance measuring matric like accuracy, recall, precision and specificity. The main focus of this research is to improve the overall performance of instructor. In this research paper C5.0 algorithm is used to predict the performance of the instructor.
Asanbe et al. 20 present a model for the evaluation and prediction about the teacher’s performance in higher institutions using data mining techniques. In this research he uses six attributes to calculate the instructor performance and analyze that the rank and experience are the two best attributes that is used for the prediction about the teacher’s performance. In this research he uses C4.5 algorithm with two other algorithms (ID3 and MLP) and calculate the 83.5% of accuracy. It does not mean that C4.5 is the best algorithm for this purpose many other algorithms are also used to produce better results.

Ughade et al. 10 presents a research paper on the performance of faculty members. In this research paper two things are calculated one is the teacher’s performance using student’s feedback and the other is result of student in that subject. In this research he use K nearest neighbor algorithm for accuracy and analyze that K nearest neighbor is the best algorithm for this kind of data set.
T.S Ha:
T.S Ha et al. 21 implemented a system for online evaluation of teaching. The data is collected from the students through the web application. In the age of technology the Questionnaire method is not very effective for this purpose so students uses website for this purpose.
Jinqiu et al. 22 presents a system for the evaluation of teacher’s performance using Analytic Hierarchy Process (AHP). AHP uses multi objectives for the decision making analysis method. This technique mainly focuses on the factors that are mainly affected on the teacher’s performance. This method is flexible but still there are many limitations in this method.
From this related work it is inferred that teacher’s performance can be analyze and improve using different data mining techniques.
In first part of this section we discuss about different methods which are used for collecting the data about the teacher’s performance and in the second part of this section we discuss about the techniques which are used to analyze and improve instructor’s performance.
A. Data collection techniques
To collect the data about the teacher’s performance different methods are used. We will discuss some methods about data collection in this section.
Questionnaire Method:
In this method different kind of questions about the teacher performance and about the subject are asked from the students and get the student’s perception. Mostly the questions are type on the word page and set the specific format of every question. The pages are given to the students and students fill the questionnaire according to their perception about the teacher and the course.

Fig 1: Ref 9

Response value of these questions are in the form {1, 2, 3, 4, 5} where 1 represents the answers “Never”, 2 represents the answers “Rarely”, 3 represents the answers “Sometimes”, 4 represents the answers “Often” and 5 represents the answers “Always” and also in the form 1, 2, 3, 4, 5 where 1 represents “Strongly disagree”, 2 represents “Disagree”, 3 represents “Neutral”, 4 represents “Agree”, and 5 represents “Strongly agree”.

Web Based Feedback Method:
In order to achieve generic behavior of online teacher feedback system use case diagram, sequence diagram and activity diagram are built to visualize the proposed system. In this proposed system web application is connected through centralized database server. In this system five peoples are directly connected with the application to make the system more reliable and efficient. First one is the system administrative that is responsible of managing the data on server. Second is the layout designer that will generate feedback layout according to the institution standard feedback format. Survey organizer will create survey by adding questions according to the layout. Students first register in the system and then perform a feedback step. Dean of the institution will see the initial feedback and then make and CSV file and send to the data mining algorithm to perform a teacher’s evaluation step.
There are various types of algorithms are used for this purpose but in this research we use Weka as a tool and J48 and BF tree decision tree algorithms are used in the given data set.

Fig 2-3: Ref 12
Opinion Mining Method:

One of the most important methods to collect the data about the instructor performance is the opinion mining. Opinion mining is most suitable method for collecting data because in this method student is not restricted with questions and he clearly write the opinion about the teacher. When the data is collected with the help of student’s opinion then a mining process is apply on the data and we predict the performance of instructor with the help of data mining techniques.
Some attributes and their values are defined in this method and the performance of teacher is predicted on the basis of these attributes. Opinion mining is not use only from the student’s side but also use the opinion of the teachers about their Faculty members. With the help of opinion the

Mining Dean of the universities and managerial peoples also predict the behavior of the teacher with their faculty members. The performance of teacher is not only about the student’s point of view but the performance of instructor is also the behavior of teacher with their colleagues and the behavior of teacher with their seniors and other high rank officers.
In this method both objective and subjective opinion are collected from student’s side but in this technique our main focus in on the subjective type of questions. In this technique we can manage both positive and negative opinions from the students. All types of questions are arranged in manners and then apply different kinds of data. Mining techniques on the data set to predict the performance of the instructor.

Fig 4: Ref 13
B. Classification Algorithms:
In this section we will discuss about some performance measuring classification algorithms and techniques that are used in educational data mining domain. There are different types of algorithms are used for this purpose but we can use only those algorithms that provide best result in a particular situation.

Decision Tree Algorithms:
Decision tree algorithm is a supervised learning algorithm that is use to solve the classification and regression problems. The main focus in this algorithm is to make a decision tree using different types of attributes. The most important attribute in the data is become the root node of the decision tree and other attributes are connected with the root node in a respective manner. The last nodes of the decision tree called the leaf nodes. Every node of the decision tree represents an attribute.

Fig 5: Ref 9

Algorithm for Decision tree is:

• Place the highest information attribute of the dataset at the root node of the tree.
• Divide the training set into subsets. Subsets should be in the form that each subset contains data with the same value for an attribute.
• Repeat above two steps on each data subset until we find leaf nodes in every branch of the tree.

Decision tree algorithms are ID3, C4.5 and C5.0 are most commonly used for classification of data into different class labels. For the selection of attributes we use some method and check that which attribute is the most important for us. For this purpose “information” Gain and “Gini index” is used.
All the equations (1), (2), and (3) are used for calculating the Information Gain, which is the impurity measure of ID3:

????(?)=????log2(??)??=1 (1)
?????(?)=?|??||?|??=1×????(??) (2)
????(?)=????(?)??????(?) (3)

Rule Based Algorithms:
Rule based algorithms are used to predict the decision about the data set. With the help of decision tree some rules are built and predict the instructor performance using these rules. In this technique the most important attribute is selected as the main rule and all the other rules are extracted with the help of this attribute.
Rule based algorithms clearly defined all the steps and these are very easy to understand the performance of the instructor.

Fig 6: Ref 13

These rules give us the clear information about the performance of the instructor that if the performance feedback and remarks are greater than 90 then the performance of the teacher is excellent and if the score of the feedback is below 90 than the performance is good and all the rules apply on the dataset in a similar manner and at the last if the feedback score is below 27 than the performance of the instructor is not satisfactory. So rule based algorithms are also very helpful for us to understand about the performance of the instructor.

Naïve Bayes Algorithm:
Naïve bayes is based on bayes theorem and in this technique the presences of one feature in the dataset is not depend on the presence of other features in the same data set. This algorithm is very helpful to experiment on a large dataset and it is also very easily implemented.
The expression used for naïve bayes is:

P(C|X) = P(X|C)* P (C) / P(X)

• P(C|X) is the posterior probability of class (target) given predictor (attribute).
• P(C) is the prior probability of class.
• P(X|C) is the ratio which is the probability of predictor given class.
• P(X) is the prior probability of predictor.

Step 1: Convert the dataset into a frequency table
Step 2: Create Likelihood table by finding the probabilities
Applying these steps repeatedly we will create a frequency table from the data set and analyze the results on the basis of frequency table.

Fig 7: Ref 13
Data mining techniques are very helpful tool to finding the hidden patterns within the educational data. If we find the hidden patterns then it is very useful activity for us because using the results of these techniques we will improve our educational system and also managers easily take good decisions. In this paper we discuss about two things one is the data collection methods about the instructor’s performance and the second thing is about the techniques which we are used to predict the instructor’s performance. These methods and techniques may helpful for the researchers to do more work in that domain.

There is lot of work that can be done in that area. Opinion mining is the most suitable option for the future work. Using opinion mining method we can easily collect the data about teacher and apply different classification algorithms I.e. Naïve Bayes, support vector machine and decision tree algorithms and easily measure the performance of the instructor in a course.

1 A. M. Abaidullah, N. Ahmed, and E. Ali, “Identifying hidden patterns in students’ feedback through cluster analysis,” International Journal of Computer Theory and Engineering, vol. 7, no. 1, pp. 16-20, 2015.

2 N. Delavari, S. Phon-Amnuaisuk, and M. R. Beikzadeh, “Data mining application in higher learning institutions,” Informatics in Education, vol. 7, no 1, pp. 31-54, 2007.

3 M. Goyal, and R. Vohra, “Applications of data mining in higher education,” International Journal of Computer Science Issue, vol. 9, no. 2, 113-120, 2012.
4 J. Luan, Data Mining and Its Applications in Higher Education in New Directions for Institutional Research, no. 113, Wiley Periodicals, Inc., 2002.

5 J. Han, M. Kamber, and J. Pei, Data mining concepts and techniques, Waltham, MA: Morgan Kaufmann Publishers, 2012.

6 J. Luan, “Data mining and knowledge management in higher education – potential applications,” In Proceedings of AIR Forum, Toronto, Canada, 2002.

7 M. B. Behrouz, D. A. Kashy, G. Kortemeyer, and W. Punch. “Predicting performance of student: a data mining technique with an educational web-based system,” In Frontiers in education, 2003. FIE 2003 33rd annual, IEEE, vol. 1, pp. T2A-13, 2003.

8 Lawrance, R., Shanmugarajeshwari, V., “Analysis of Students’ Performance Evaluation using Classification Techniques.” IEEE International Conference on Computing Technologies and Intelligent Data Engineering (ICCTIDE’16), 10.1109/ICCTIDE.2016.7725375, 16426971

9 Agaoglu, M., IEEE Translation and Content Mining, “Predicting Instructor Performance Using Data Mining Techniques in Higher Education.” Department of Computer Engineering, Marmara University, Istanbul 334722, Turkey. Volume.4, 2016, pp:2379-2387

10Ughade, P., Mohod, W.S., “A Survey on Analysis of Faculty Performance Using Data and Opinion Mining.” International Journal of Innovation Research in Computer and Communication Engineering (IJIRCCE), Volume. 4, Issue. I, January 2016, pp: 87-91

11 D.C.V, “Evaluation of teacher’s performance using fuzzy inference techniques “, international Journal of computer trends and technology, 2012, vol 3(2), pp 34-40.

12 J. Sang, H. Naeem (2015). A Generic Feedback System for Better Evaluation of Teacher Performance.

13 S. Kim and E. Hovy (2004). “Determining the Sentiment of Opinions”, Proceedings of the COLING conference, Geneva, 2004

14 Feinerer, I., Hornik, K., ; Meyer, D. (2008). Text Mining Infrastructurein R. Journal Of Statistical Software, 25(5),1–54.Available at

15 V. Kumar, and A. Chadha, “An empirical study of the applications of data mining techniques in higher education,” International Journal of Advanced Computer Science and Applications, vol. 2, no. 3, 80-84, 2011.

16 C. Romero, and S. Ventura, “(EDM) Educational data mining: A survey ranges from 1995 to 2005,” Expert Systems with Applications, vol. 33, pp. 135-146, 2007.

17 B. K. Baradwaj, and S. Pal, “Mining Educational Data to Analyze Students’ Performance,” International Journal of Advanced Computer Science and Applications, vol. 2, no. 6, pp. 63-69, 2011.

18 S. Calkins, and M. Micari, “Less-than-perfect judges: evaluating student evaluations,” The NEA Higher Education Journal, pp. 7-22, Fall, 2010.

19 J. Sojka, A.K. Gupta, D.R. Deeter, “Perceptions of students and faculty members of student evaluations of teaching: a study of similarities and differences,” College Teaching, vol. 20. No. 2, pp. 44-49, 2002.

20 Asanbe, O.M., Osofisan, O.A., and William, F.W., “Teachers’ Performance Evaluation in Higher Educational Institution Using Data Mining Technique.” International Journals of Applied Information System (IJAIS) Volume. 10, No. 7, March 2016, pp:10-15

21 T. S. Ha. J. Marsh, J. Jones, ” A Web-Based system for teaching evaluation ” Lingnam college 30th anniversary conference, New Challenges and Innovations in Teaching and Training into the 21th century (NCITT’98) , pp. 1-11 May 1998.

22 L. jinqiu, Liang L. J. Liang, L. Huimin, “AHP-Based evaluation on college teacher’s classroom training quality” IEEE international conference on business management and electronic information (BMEI), pp. 558-560. May 2011.

You Might Also Like

I'm Alejandro!

Would you like to get a custom essay? How about receiving a customized one?

Check it out