MARYLI F. ROSAS et al: EXTRACTION OF STUDENTS’ FEEDBACK IN UNIVERSITY SERVICES USING . . DOI 10.5013/IJSSST.a.20.02.08 8.1 ISSN: 1473-804x online, 1473-8031 print Extraction of Students’ Feedback in University Services using Logistic Regression and Chi-Square Automatic Interaction Detection (CHAID) Algorithm Maryli F. Rosas 1 , Shaneth C. Ambat 2 , Ace C. Lagman 3 1 Computer Studies Department, De La Salle University – Dasmariñas, City of Dasmariñas, Philippines. 2 Graduate School, FEU Institute of Technology, Manila, Philippines. 3 Information Technology, FEU Institute of Technology, Manila, Philippines. Email: mfrosas@dlsud.edu.ph, shaneth_ambat@yahoo.com, aclagman@feutech.edu.ph Abstract - The ultimate goal of every Higher Education Institution is to provide quality education to their students. Quality of Education can be measured through the student’s satisfaction level based on their overall university experience. Students are the primary consumers in every Higher Education Institution services. Hence, their feedback or insights on their overall university experiences are significant for the university’s continuous quality improvement. The use of machine learning algorithm concepts can determine and extract useful information which can be used as reference in formulating necessary policies suited to improve university’s academic services. Our study focuses on the application of logistic regression through equation development and CHAID Algorithm through rules development. The study also uses text analytics concepts in which sentiment scores are computed to classify the comments according to satisfaction level. Key words - Data Mining, CHAID, regression, sentiment, text analytics, University Services A shorter version of this paper with title “Data Mining of Students’ Response in the University Services using Chi-Square Automatic Interaction Detection (CHAID) was presented at 1st IEEE International Conference on Knowledge Innovation and Invention (ICKII), July 23 to July 27, 2018, Jeju Island Korea[2]. I. INTRODUCTION Higher Education targets to develop complex theoretical, abstract and analytical reasoning capabilities in the alumni [3]. Quality education can be determined thru the quality of services that were given to the students. It is every organization’s effort to continuously improve their quality of services. In order to measure students' reactions, evaluation forms can be used by educational institutions [4]. In the university setting, it is very vital for them to know their performance through gaining insights from the students in a form of an exit interview. Data mining in education enables data-driven decision-making for improving current educational practice and learning materials [5]. Another paper justified the capabilities of data mining techniques in context of higher education by offering a data mining model for higher education system [6]. The general objective of this study is to develop and implement an improvement plan based on the significant patterns produced by the CHAID algorithm to improve the quality of services of the university. This study could be a significant endeavor in monitoring and escalating the effectiveness and efficiency of the units contributing to the over-all quality of student services. The study investigated on how students' experiences affect their satisfaction level by using data mining techniques particularly the implementation of regression and decision tree analysis with Chi-square Automatic Interaction Detector (CHAID) algorithm. Moreover, this study implemented Sentiment Classification Technique using supervised Machine Learning approach [7] that will correlate to CHAID algorithm. This paper would like to address the following research questions: 1. How logistic regression determines significant attributes in students satisfaction towards academic services? 2. How to extract predictive data models using Decision Tree Algorithm under Chi-square Automatic Interaction Detector (CHAID method? 3. How acceptable the software developed using software evaluation instrument as perceived by IT Expert in terms of the following criteria a. Functionality b. Usability c. Reliability d. Performance e. Security It is worth noting that on the study of student satisfaction is not yet fully utilized. And the implication of the result of such would help the Higher Education Institution to improve on their quality of education. Hence, the researcher is motivated to explore and discover more of this area. The researchers used Knowledge Discovery in Database in which machine learning algorithm was applied to generate model. In this study, the researcher used logistic regression analysis to determine the significant attributes. Through executing Logistic Regression analysis model, the Sentiment,