Mogućnosti primene poslovne inteligencije za analizu i predviđanje uspeha studiranja

Abstract

Kontinualno obrazovanje i stalno usavršavanje predstavlja jednu od osnovnih paradigmi napretka savremenog društva, zasnovanog na tehničko-tehnološkom razvoju i globalnom poslovanju. Obrazovanje i sticanje znanja danas postaje mnogo brže, praćeno informatičkom evolucijom i napredovanjem tehnologija, što dovodi do novih naučnih saznanja i primena, ali i razvoja novih naučnih oblasti. Početkom XXI veka, nastala je jedna nova naučna oblast, kao disciplina poslovne inteligencije, za otkrivanje zakonitosti u podacima iz oblasti edukacije (eng. Educational Data Mining), koja se bavi razvojem metoda za istraživanje i utvrđivanje zakonitosti u podacima koji dolaze iz oblasti edukacije i koristi svoje metode radi boljeg razumevanja ponašanja studenata i realizacije nastavnog procesa. Predmet doktorske disertacije predstavlja mogućnost primene poslovne inteligencije u obrаzovаnju, kаko bi se identifikovali ključni činioci uspeha studenata radi poboljšanja visokoškolskog obrazovnog procesa. Utvrđivanje najznačajnijih varijabli o studentu, od sredine iz koje dolazi i završene srednje škole, uspeha na prvoj godini osnovnih akademskih studija, preko njihovog uticaja i korelacije sa celokupnim uspehom studiranja, do mogućnosti preciznog predviđanja uspeha na kraju studija, primenom metoda, tehnika i alata poslovne inteligencije, predstavlja centralni predmet istraživanja disertacije. Doktorska disertacija detaljno analizira dosadašnja istraživanja i saznanja iz oblasti otkrivanja zakonitosti u podacima iz oblasti edukacije, a posebno visokoškolske edukacije, na osnovu koga je uočena potreba za predlogom novog pristupa i metoda za predviđanje indikatora uspešnosti studiranja na visokoškolskim institucijama. Disertacija takođe daje predlog originalnog softverskog rešenja–aplikacije za analizu, praćenje i predviđanje uspeha studenata na osnovnim akademskim studijama, kao deo integrisanog portala za razmenu znanja i informacija na relaciji student-fakultet. Primenom metoda i tehnika poslovne inteligencije, izvršena je implementacija razvijenih modela za predviđanje uspeha studiranja, kao i verifikacija dobijenih rezultata nad bazom studenata Fakulteta organizacionih nauka, Univerziteta u Beogradu. Doktorska disertacija je strukturirana u 12 poglavlja, gde su u uvodnom poglavlju predstavljeni ciljevi, hipoteze, kao i plan realizacije i istraživanja doktorske disertacije. Potom su kroz naredna dva poglavlja dati osnovni koncepti poslovne inteligencije i otkrivanja zakonitosti u podacima, uključujući analizu i sistematizaciju saznanja istraživanjem obimne literature o otkrivanju zakonitosti u podacima u oblasti visokoškolske edukacije. Četvrto poglavlje daje osnovni prikaz projektovanog modela istraživanja, pre svega u pogledu korišćene softverske podrške za realizaciju istraživanja, kao i izvora i strukture podataka koji su korišćeni. U petom poglavlju su primenjeni koncepti deskriptivne i komparativne analize, za detaljnije utvrđivanje strukture podataka, međusobnih zavisnosti i korelacija unutar podataka.

Lifelong education and constant training, represents one of the basic paradigms of progress of modern society, based on the technical-technological development and global business. Education and learning has become much faster, followed by IT evolution and technology progress, leading to new scientific knowledge and application, but also the development of new scientific fields. At the beginning of the XXI century, there was a new area of science, as a discipline of business intelligence, to discover the relations of the data in the field of education - Educational Data Mining, which deals with the development of methods for analyzing and determination of the legality of the data coming from the field of education and uses its methods to better understand the behavior of students and realization of teaching progress. The subject of the PhD thesis presents the possibility of applying business intelligence in education, in order to identify the key factors of student’s success in order to improve the higher education process. Determining the most important variables of a student from his/her origin and high school, success in the first year of undergraduate studies, through their influence and the correlation with the overall success of the studies, to the ability to accurately predict success at the end of the studies, using methods, techniques and tools of the business intelligence, is the central subject of dissertation. PhD thesis analyzes current research and knowledge in the field of data mining in education in detail, especially in the field of higher education, on the basis of which the need for proposing a new approach and method for predicting performance indicators of study in higher education institutions is perceived. Dissertation, also proposes original software solution–application for analysis, monitoring and forecasting the success of students at the undergraduate level, as part of an integrated portal for the exchange of knowledge and information between student and faculty. By applying the methods and techniques of the business intelligence, the implementation of the developed models to predict the success of the study was made, as well as the result verification of the data base of students of the Faculty of Organizational Sciences, University of Belgrade. PhD dissertation is structured in 12 chapters. Objectives, hypotheses, as well as the plan for the implementation of research and doctoral dissertations are being presented in the introductory chapter. Then the next two chapters provide the basic concepts of business intelligence and data mining, including the analysis and systematization of knowledge and extensive research literature on data mining in the field of higher education. The fourth chapter provides a basic outline of the projected research model, especially in terms of software support used for the implementation of research, as well as sources and data structures that are used. In the fifth chapter concepts of descriptive and comparative analysis are being applied, for a detailed definition of data structures, mutual dependencies and correlations within the data.