Analyzing Student Evaluations of Teaching in a Completely Online Environment

Student evaluations of teaching, also known as SET, is an integral aspect of universities' programs for selfimprovement. SET has three primary purposes: (a) to improve teaching quality, (b) to inform tenure/promotion choices, and (c) to demonstrate an institution's accountability [6]. Improving teaching quality is a major academic objective for all universities. Most people see the university as just another business, and every company's primary goal is to meet the requirements of its clients. In this instance, university clients are students. Because of this, it is important to think about how students feel in order to improve the quality of teaching. At the end of semester SET are typically given out anonymously. Students are asked to evaluate various aspects of the course and the teacher on a Likert scale. In addition to specific course details, such as course structure or grading policies, many SET give an overall assessment of the teacher and the course. SET may also include open-ended questions, in which students are invited to provide opinions on the course and the teaching methods of the teacher [7].

The aim of this study is to provide an extensive analysis of student evaluations of teaching in a completely online environment by using educational data mining. Educational Data Mining (EDM) is a technique used by educational institutions to locate usable data that can aid in the improvement of their educational activities. Thirty-five course evaluations of teaching were gathered from three fully online semesters to analyze, where computer science students at X University were participated anonymously. This study will assist in identifying the factors that play a significant role in improving teaching quality. If the need arises to move from an online to a traditional classroom setting, this research can serve as a guide for improving teaching quality. Moreover, it can also be used as a guide to make major educational decisions, particularly for computing students.

The study is comprised of five sections, which are structured as follows, an introduction is found in section 1, followed by a summary of the overall situation of student evaluations of teaching in section 2. The methodology is stated in section 3. Section 4 includes the analysis of the obtained data and findings. Lastly, section 5 consists of the conclusion and discussion.

2.    Background Study 2.1.    Classification and analysis of Likert scale questions

The educational environment often uses data mining techniques to explore and evaluate teaching performance in order to improve the quality of teaching. There are various educational data mining techniques such as Naive Bayes, K-Nearest Neighbors, Decision Tree, Association Rules Mining, Regression, and many more for identifying the main factors of SET and classifying the courses based on the rating of SET. This study [7] employed decision tree analysis to predict high overall teaching and course scores on a SET instrument. There were 98,525 SET records for 2,870 instructors across 5,076 course sections. Both the “teaching effectiveness” decision tree and the “course effectiveness” tree utilized a total of 96,385 and 95,780 SETs, respectively. SET was comprised of 16 evaluation elements. Results showed that helping students grasp course material and offering an intellectually exciting class were most predictive of a student's overall teacher evaluation. On the other hand, helping the student better grasp the course material and assignments that helped students learn the topic were the two factors that were most predictive of a student's overall rating of the course. This paper [8] applied data mining to evaluate teaching quality. As for test data, they extracted 3000 records, which included information on teachers' qualities, teaching duties, and evaluations. They constructed two data mining models (Association Analysis and Decision Tree) to describe how instructor and course quality affected evaluations. The investigated data revealed several hidden correlations between the teacher's features and the results of in-class teaching quality evaluations. They also showed how course features such as property, credit, week-hour, and number of students affected teaching quality evaluation outcomes. The goal of this study [9] was to make and test a measuring tool that could be used to rate university professors. The original sample contained a total of 1297 students' ratings. Using the Evaluation of Teaching Performance (ETP) questionnaire, 28 items were categorized into three factors: planning; development; and result. Items were scored on a Likert scale from 1 to 5. These three factors had an average value of 3.95, 3.77, and 3.74, respectively. This study [10] aims to examine the elements affecting student views of instructional effectiveness. The data came from a Course Evaluation Survey (CES) where around 3798 students responded where 2159 were female and 1639 were male. The survey was comprised of 26 specific questions regrouped under five factors, such as the instructor's personality characteristics, their behavior in marking and grading, their knowledge and teaching skills, the course attributes, and the course learning outcomes. Results showed a strong association between student evaluation of teaching effectiveness and each of the five criteria. The instructor's personality was the most important factor for both male and female students. Then came the course's characteristics, the instructor's knowledge and teaching skills, the course's learning outcomes, and finally, the instructor's methods for marking and grading. In the article [11], the impacts of perceived course efficiency on multiple SET measures were studied. The data came from 280 students enrolled in 14 sections of three online marketing courses, i.e., Marketing Research, Marketing of Financial Services, and Marketing Strategy. All the indicators of SET were assessed on a 5-point Likert scale. Regression analysis found that students who rated an online course as more effective than a face-to-face course had a weaker academic profile. The findings also revealed that students rated their instructors and courses, on average, 4.67 and 4.43, respectively. Most students who took online marketing courses considered them to be as effective as face-to-face courses. Overall, the Marketing Strategy course was rated higher than the other two courses. In the study [6], it examined students' perceptions of SET practice and their SET results. There were 974 students who participated in that study. Participants completed a SET and an adapted version of the Students' Perceptions of a Teaching Evaluation Process Questionnaire, which had 15 items scored. SET was viewed positively by nearly 90% of respondents, who believed it could improve teacher quality. Participants agreed that senior faculty shouldn't be judged as heavily as junior faculty. The authors also identified a strong association between SET scores and student perceptions of SET practice. Both SET scores and how important students thought SET practice was were related to their grades, seniority, and academic discipline. The study [12] sought to examine medical students' course evaluations in order to enhance the course evaluation system. 93 medical students responded to a questionnaire designed for computer-aided course evaluation. The questionnaire contained a total of 21 different assessment questions. The overall percentage of students who responded was 63.8% on average. Responses were more negative in 17 Likert-scaled categories, such as class size, suggested ways to study, lecture notes, teaching methods, and how important the final exam was.

• 2.2.    Classification and analysis of opinions

• 2.3.    Selection of classifier and opinion analysis technique

3.    Methodology 4.    Data Analysis and Findings 4.1.    Data Collection and Preprocessing

Many teachers' evaluation forms include a part where students are free to express their opinions about the course and teacher. Various classification and analysis techniques are utilized to determine these opinions. The article [13] proposes two lexical-based techniques for automatically extracting opinions from brief reviews. There were 3,926 SET records that featured short review texts. They received 1,323 texts expressing negative emotions and 1,699 texts expressing positive emotions. Four machine learning approaches (Naive Bayes, Logistic Regression, Support Vector Machine, and Gradient Boost Decision Tree) were employed to calculate the emotional polarity of the texts. The proposed approaches were found to be 78.13% and 84.78% accurate at classifying the sentiment of student reviews for positive and negative emotions, respectively. In the study [14], an opinion mining method was presented to improve course evaluations. Data was gathered through the use of 4,957 discussion postings from three different discussion boards. For the purpose of opinion classification, three different machine learning approaches (Naive Bayes, KNN, and Support Vector Machine) were utilized. According to the results, the Naive Bayes approach performed better than the other two methods. Extracting opinions revealed that the characteristics with the highest frequency were teachers, content, exams, marks, and books. The authors of this research [15] employed Valence Aware Dictionary and sEntiment Reasoner (VADER) to examine Student Evaluations of Teaching (SET) from three sources, i.e., official evaluations, forum comments from another course, and an unofficial “ reviews ” site maintained by the students. Results showed that all three sources scored nearly the same on compound ratings, although unofficial student site ratings were somewhat more negative and included fewer favorable responses. The average rating for those who gave 5 stars was nearly completely positive, at 97.96%. Overall, questions about the strengths of the instructor or the course got the fewest negative and most positive scores, and questions about their weaknesses got the opposite. Furthermore, many scholars use Natural Language Processing (NLP) techniques to analyze opinions on various platforms. The authors of [16, 17, 18, 19, 20, 21, 22, 23] used sentimental analysis based on NLP on Twitter data to figure out whether a tweet is positive, negative, or neutral. Also, sentimental analysis was used in many different fields, such as Twitter tweets about COVID-19 vaccines [24], social media based on COVID-19 [25], US airline Twitter data [26], healthcare [27], movie reviews [28], product reviews [29, 30, 31], and so on.

A decision tree is likely the optimal starting point for classifying any dataset. It will provide an excellent overview and make the classification clear. Decision trees contain straightforward characteristics for defining the most important dimensions, managing missing values, and handling outliers. The majority of individuals are familiar with hierarchical trees; therefore, a basic graphic will facilitate the communication of the findings. Decision trees are straightforward to comprehend. People can obtain a general idea of the tree simply by glancing at it. Additionally, [7, 8] employed a decision tree to identify the most essential SET factors. Consequently, earlier scholars have solved a variety of classification issues with decision trees. Besides, scholars frequently use sentiment analysis while assessing opinions, as we have already mentioned above. As a result, the decision tree classifier and sentimental analysis will be used to carry out this study.

In the below fig. 1 shows the experimental flowchart of our SET data analysis. Data is collected from X University, which contains 27,622 SET records of 701 sections for 35 different courses. Then we do preprocessing in order to apply the data mining algorithm. After that, opinions are analyzed. Lastly, all the findings are gathered for discussion.

The SET dataset has the numeric values for five different factors. Here, we are using Weka where k-means clustering will be used for grouping the data of class level and, lastly, a decision tree (J48) will be generated to identify the prominent factors that have significant impacts on SET. There are also qualitative opinions that are difficult to summarize in SET dataset. That's why Sentimental Analysis, a type of natural language processing, will be utilized to determine the polarity and subjectivity of the opinions. This will assist in identifying whether the opinion is positive, negative, or neutral. The Word Art online-based word cloud generator website is also going to be used here to find the most frequently used words in the opinions. However, the flowchart of the proposed work in given below.

Fig. 1. Flowchart of the SET data analysis

Dataset is collected from X University's SET reports of computing students in three different online semesters. This data contains a total of 27,622 SET records of 701 sections for 35 different courses. Each section is filled with 36 to 42 students. Every student participated in these SET through the university portal at the end of the semester. Table 1 demonstrates the sample size of the students and sections for 35 different courses offered in the back-to-back three online semesters.

Table 1. Total SET dataset

Semester	Students	Sections	Total Courses
Spring 2020-21	8595	209	35
Summer 2020-21	9373	232
Fall 2021-22	9654	260
Total	27622	701

Each set of SET consists of a total of twenty questions. These twenty questions are designed to measure five factors of teaching quality, where each factor contains four questions. The Likert scale is used to provide a score for each question ranges from one to five. The factor names are shown in below table 2.

Table 2. Factors of SET

Factor SL.	Factor Name
Factor 1	Knowledge of the Subject Matter
Factor 2	Instructional Strategies and Motivation Techniques
Factor 3	Personality Traits
Factor 4	Student-Faculty Relation
Factor 5	Routine Matters

A factor-wise rating is calculated from the average point of its four questions. An overall rating is calculated from the average of these five factors. There is also a section at the end of SET for giving opinions about the course and teacher. The SET questioner is provided in the Appendix A section. The names of the courses and their tag numbers are given in below table 3.

Table 3. List of courses and their tag numbers

Course Tag	Course Name
C1	Introduction to Computer Studies
C2	Introduction to Programming
C3	Introduction to Programming Lab
C4	Discrete Mathematics
C5	Object Oriented Programming 1
C6	Introduction to Database
C7	Data Structure
C8	Algorithms
C9	Object Oriented Programming 2
C10	Object Oriented Analysis and Design
C11	Theory of Computation
C12	Data Communication
C13	Software Engineering
C14	Artificial Intelligence and Expert System
C15	Computer Networks
C16	Computer Organization and Architecture
C17	Operating System
C18	Web Technologies
C19	Compiler Design
C20	Computer Graphics
C21	Research Methodology
C22	Advance Database Management System
C23	Data Warehousing and Data Mining
C24	Human Computer Interaction
C25	Software Development Project Management
C26	Software Requirement Engineering
C27	Software Quality and Testing
C28	Programming in Python
C29	Advanced Programming with .Net
C30	Advanced Programming in Web Technology
C31	CS Math
C32	Basic Graph Theory
C33	Advanced Operating System
C34	Computer Vision and Pattern Recognition
C35	Network Security

As mentioned earlier, the dataset contains a total of 701 sections for 35 different courses. Based on their overall SET score, these courses are categorized into three groups by using k-means clustering algorithm and Weka tool. We named these three groups as very effective course, effective course, and course needs to improve, as shown in the table 4 below.

Table 4. Class level of courses by using k-means clustering algorithm

Class Level	SET Overall Rating	Course Count
Very effective course	4.58 to 5	1
Effective course	4.17 to 4.57	18
Course needs to improve	Less than 4.17	16

Based on the overall rating of SET, three categories of courses are found, where one course is very effective, eighteen courses are effective, and sixteen courses are identified for improvement. Class level-wise course tags are provided below in below table 5, and fig. 2 shows the percentage distribution of courses’ classification.

Table 5. Group of courses based on overall rating of SET

Class Level	Course Tag
Very effective course (4.58 <= Overall Rating <= 5)	C24
Effective Course (4.17 <= Overall Rating <= 4.57	C1, C2, C3, C4, C5, C6, C9, C16, C21, C22, C23, C25, C27, C29, C30, C32, C34, C35
Course needs to improve (Overall Rating < 4.17)	C7, C8, C10, C11, C12, C13, C14, C15, C17, C18, C19, C20, C26, C28, C31, C33

Fig. 2. Percentage distribution of courses’ classification

• 4.2.    Implementation Decision Tree Classifier

For identifying the most influential factor among those five factors of SET, course names and tags are ignored. Therefore, it will now be an overall semester's SET records. Weka, a collection tool of machine learning algorithms for data mining tasks [32], is used to apply the J48 decision tree algorithm in 10-fold cross-validation settings.

The outcome of the J48 decision tree algorithm is shown below in fig. 3, and the rules are also shown in table 6. Generally, the class levels are stated in the leaf node of the decision tree. We can see from the decision tree in fig. 3 that factor 4 has the most influence because it is the tree's root node. That means, if you want to reach any class level, first you must meet the condition of the rating for factor 4.

Fig. 3. Decision tree of the SET dataset generated by Weka

Table 6. Derived rules of the decision tree

Rules	Description of the rules
R1	If [Factor 4 > 4.15 && Factor 3 > 4.57 && Factor 4 > 4.55] then Class level = ‘Very effective course’
R2	If [Factor 4 > 4.15 && Factor 3 > 4.57 && Factor 4 <= 4.55] then Class level = ‘Effective course’
R3	If [Factor 4 > 4.15 && Factor 3 <= 4.57 && Factor 1 > 4.14] then Class level = ‘Effective course’
R4	If [Factor 4 > 4.15 && Factor 3 <= 4.57 && Factor 1 <= 4.14 && Factor 1 > 4.12] then Class level = ‘Effective course’
R5	If [Factor 4 > 4.15 && Factor 3 <= 4.57 && Factor 1 <= 4.14 && Factor 1 <= 4.12 && Factor 3 > 4.19] then Class level = ‘Effective course’
R6	If [Factor 4 > 4.15 && Factor 3 <= 4.57 && Factor 1 <= 4.14 && Factor 1 <= 4.12 && Factor 3 <= 4.19] then Class level = ‘Course needs to improve
R7	If [Factor 4 <= 4.15 && Factor 3 > 4.12 && Factor 2 > 4.18] then Course level = ‘Effective course’
R8	If [Factor 4 <= 4.15 && Factor 3 > 4.12 && Factor 2 <= 4.18 && Factor 5 > 4.21 && Factor 3 > 4.15] then Class level = ‘Effective course’
R9	If [Factor 4 <= 4.15 && Factor 3 > 4.12 && Factor 2 <= 4.18 && Factor 5 > 4.21 && Factor 3 <= 4.15] then Class level = ‘Course needs to improve’
R10	If [Factor 4 <= 4.15 && Factor 3 > 4.12 && Factor 2 <= 4.18 && Factor 5 <= 4.21] then Class level = ‘Course needs to improve’
R11	If [Factor 4 <= 4.15 && Factor 3 <= 4.12] then Class level = ‘Course needs to improve’

The performance of the decision tree algorithm is shown in table 7 below. In Weka, the accuracy of the model is determined by the percentage of instances that were properly classified. The Kappa statistic is also given here, which is a measurement that compares an observed accuracy to an expected accuracy. Therefore, the values of True Positive (TP) Rate, False Positive (FP) Rate, Precision, Recall, and F-Measure for all possible class levels are given here. False Negative (FN) Rate and True Negative (TN) Rate are not included as these two are residual values of 1 for TP Rate and FP Rate, respectively. Here, the decision tree has a very high degree of accuracy, which is around 96%.

Table 7. Performance indicators of the decision tree

	Possible levels of class attribute
	Very effective course	Effective course	Course needs to improve
TP Rate	0.864	0.965	0.967
FP Rate	0.005	0.050	0.023
Precision	0.944	0.948	0.970
Recall	0.864	0.965	0.967
F-Measure	0.903	0.956	0.969
Total number of instances	701
Correctly classified instances	671
Incorrectly classified instances	30
Accuracy	95.7204%
Kappa statistic	0.9248

• 4.3.    Implementation of Opinion Analysis

Opinions are classified using sentimental analysis. This will aid in determining whether the opinion is favorable, unfavorable, or neutral. With the help of four built-in libraries (pandas, regular expression, matplotlib, and TextBlob) in Python, the opinions are classified. Basically, opinions are grouped based on their polarity score. If the polarity score of an opinion is zero, it is treated as a neutral opinion. Positive and negative opinions are found if the score of the polarity is greater than zero and less than zero, respectively.

The result of the sentimental analysis for each course is shown in below table 8, where course tag 32 has the most positive opinions and no negative opinions, course tag 26 has the least positive opinions, and course tag 17 got the most negative opinions.

Table 8. Course-wise opinion analysis

Course Tag	Opinion Analysis
	Positive	Neutral	Negative
C1	81.03%	14.51%	4.46%
C2	81.71%	13.86%	4.43%
C3	79.64%	16.29%	4.07%
C4	81.39%	14.73%	3.88%
C5	83.10%	15.13%	1.77%
C6	79.58%	18.46%	1.96%
C7	78.46%	18.23%	3.31%
C8	77.03%	17.91%	5.06%
C9	80.47%	16.83%	2.70%
C10	75.79%	18.86%	5.35%
C11	77.26%	19.09%	3.65%
C12	75.50%	20.50%	4.00%
C13	78.68%	18.28%	3.04%
C14	75.07%	19.66%	5.27%
C15	75.56%	19.82%	4.62%
C16	82.53%	15.96%	1.51%
C17	73.47%	18.57%	7.96%
C18	76.59%	18.97%	4.44%
C19	77.24%	19.04%	3.72%
C20	76.10%	19.32%	4.58%
C21	76.72%	19.58%	3.70%
C22	73.84%	24.42%	1.74%
C23	75.27%	21.86%	2.87%
C24	79.75%	18.77%	1.48%
C25	76.74%	20.83%	2.43%
C26	69.35%	23.64%	7.01%
C27	76.73%	19.31%	3.96%
C28	73.24%	20.42%	6.34%
C29	78.71%	20.64%	0.65%
C30	80.00%	16.80%	3.20%
C31	74.04%	19.23%	6.73%
C32	87.83%	12.17%	0.00%
C33	73.21%	22.33%	4.46%
C34	77.78%	17.59%	4.63%
C35	77.33%	19.43%	3.24%

Fig. 4 shows the “Subjectivity” and “polarity” indicators, where subjectivity indicates how subjective or opinionated the text is, and polarity indicates how positive or negative the text is. Here, a polarity score greater than zero has the highest density, which indicates that a huge number of opinions are positive.

Fig. 4. Subjectivity vs Polarity of the opinions

Fig. 5 shows the percentage distribution of positive, negative, and neutral opinions of sentimental analysis, where al-most 78% opinions are positive which indicates a very positive academic environment.

■ Positive Opinions - Neutral Opinions ■ Negative Opinions

Fig. 5. Percentage distribution of sentimental analysis of the SET dataset

Moreover, an online-based word cloud generator tool is used here to determine the most frequently used words in the opinions. Basically, a word cloud is a visual representation of a group of words in varying sizes. The bigger and bolder the word appears, the more often it’s mentioned within a given text, and the more important it is, shown in fig. 6, where positive words reflect the positive opinion percentage.

Fig. 6. Word cloud for the dataset

5.    Conclusion and Discussion

Because of the pandemic that has spread over the world, educational institutions were required to move all of their activities online. The viability of taking lessons online is currently being called into question. Aside from that, education must be maintained at its highest level by analyzing student evaluations of teaching and implementing the necessary measures for improvement. Student evaluations of teaching (SET) are an important way to find out how good a teacher is at teaching, and they have become an important part of managing higher education. According to the findings, a variety of elements influence the quality of teaching which are shown in below table 9.

Table 9. Summary of the findings

Terms	Findings
Most important factor based on DT classifier	Student-Faculty Relation
Most important factor based on average	Routine Matters
Most frequent word that appears in the opinions	Good
Very effective course based on overall rating	Human Computer Interaction
Course with most positive comments based on opinion analysis	Basic Graph Theory
Course with less negative comments based on opinion analysis
Course with most negative comments based on opinion analysis	Operating System
Course with less positive comments based on opinion analysis	Software Requirement Engineering

Student-Faculty Relation is the most important factor for determining the teaching quality because it is the root node of the tree. That means, the very effective course depends on the rating of factor 4. The accuracy for the decision tree is very good, which is above 95%. It indicates that this model nearly perfectly classifies its instances. Additionally, the kappa statistic value is greater than 0.90. When comparing observed accuracy to expected accuracy, a kappa value of more than 0.75 is usually a sign of excellent agreement. So, kappa statistics also indicate that this model is perfectly classified.

As the classes were online, teachers were easily able to take the classes virtually. They could simply check the attendance of students by clicking some buttons on their computers. However, taking the class on time and uploading study materials smoothly within a short time frame are the major advantages of taking online classes. These things are easily followed by the teacher without any hassle in virtual classes. Factor-5, known as “ Routine Matters ”, has all these rating type questions, which are shown in the Appendix A section. That is the main reason for getting the highest average rating in factor 5 by the students. But a teacher could never keep an eye on all the students in an online class. So, some communication gaps might arise between the teachers and the students, which is the major drawback of virtual classes. Students might fail to obtain sufficient information about the contents of the course owing to a lack of interaction with the teacher sometimes. That might be the reason for having bad grades in the course. As a result, students could give their SET for courses and teachers very poor ratings. The findings of the Decision Tree also reveal that “ Student-Faculty Relation ”, i.e., factor-4 is the most vital factor determining the quality of teaching and very effective courses. Therefore, maintain good student-faculty relation is the key factor in online class environment, and “ Routine Matters ”, i.e., factor-5 is also important. Teachers must maintain these factors while taking online classes in the future. In addition, the average overall ratings and sentiment analysis of the courses show that “Human Computer Interaction” is the very effective course, followed by the “Basic Graph Theory” course. Besides, “Good” is identified as the most frequent word in the opinions section on SET. The results will have a big effect on how well teachers will prepare for the future if they have to move their regular classes online. Also, teachers will be able to make better decisions regarding the quality of their teaching.

Nevertheless, this study has several drawbacks. Unfortunately, the dataset doesn’t include the gender of the students or the teachers. That is why gender bias on SET was not conducted. For a particular student, the correlation between the rating of SET and giving the levels of opinion was not demonstrated as the dataset only contains sectionwise SET ratings and opinions. However, some opinions on SET were mixed with the native and the English language with many grammatical mistakes. We had to deal with those problems manually while analyzing the opinions. SET also contained only emoji-type opinions, which were marked as neutral by sentimental analysis. These challenges can be resolved in future iterations. In the future, we will conduct a comprehensive analysis of student evaluations of teaching across four academic years' course curriculum.

Appendix A. Questions for Student Evaluations of Teaching (SET)

Factor	Questions
1.   Knowledge of the Subject Matter	1.    Shows intellectual grasp of the subject matter consistent with the course outline or syllabus. 2.   Relates subject matter to actual life experiences by providing examples, illustrations. 3.   Plans and organizes the lectures, discussions properly, logically with the help of instructional tools such as: multi-media, white board, OHP etc. 4.   Asks thought provoking questions which allow students to think critically and express ideas clearly with confidence.
2.   Instructional    Strategies    and Motivation Techniques	5.   Applies different approaches, styles and techniques in teaching and evaluating students’ performance. 6.   Shows ability to manage classroom situation such as: discipline of students, proper decorum and orderliness in the classroom. 7.   Provides additional information to support and enrich students’ presentations and discussions. 8.   Encourages students to ask questions, clarifications on the topic being discussed and acknowledges students good responses and participation.
3.   Personality Traits	9.   Shows self-confidence, tact, patience and proper behavior in dealing with the students inside and outside of the classroom. 10.    Possesses pleasing personality, good grooming and properly dressed. 11.    Shows interest and dedication to teaching. 12.    Good command of the English language and uses it as a medium of instruction.

4.   Student-Faculty Relation	13.    Maintains mutual respect between the teacher and students and among students. 14.    Provide conducive classroom atmosphere and opportunities for students to spontaneously and effectively interact with each other. 15.    Demonstrates concerns and welfare of students through appropriate counseling and assistance in their studies. 16.    Practices fairness, objectivity and rationale in taking actions for and against the students.
5.   Routine Matters	17.    Regularly check attendance of students. 18.    Punctual in reporting to the class. 19.    Upload lecture notes, assignments, case studies, exercise in the website. 20.    Policies, grading system, procedures are clearly explained and compiled with by both the students and teacher.
Comment: