Title Korištenje društvenih mreža od strane studentske populacije
Title (english) Using social networks among student population
Author Lidija Vuk
Mentor Mirjana Pejić Bach (mentor)
Committee member Jovana Zoroja (predsjednik povjerenstva)
Committee member Ivan Strugar (član povjerenstva)
Granter University of Zagreb Faculty of Economics and Business Zagreb
Defense date and country 2016-05-11, Croatia
Scientific / art field, discipline and subdiscipline SOCIAL SCIENCES Economics Business Informatics
Abstract Da bi se učinkovito iskoristilo znanje iz velikih količina podataka pomaže široki spektar
aplikacija i tehnologija za prikupljanje podataka i ekspertnu analizu, odnosno poslovna
inteligencija koju definiramo kao ranije prikriveno znanje koje se otkriva iz operativnih,
rutinski prikupljenih poslovnih podataka primjenom računsko-logičkih metoda te najčešće
koriste informacijsku tehnologiju. Velike baze podataka bogate su podacima, ali su
istovremeno siromašne informacijama. Podatak se dobiva zapažanjem ili mjerenjem dok se
davanjem konteksta podatku dobiva informacija. Znanje je potrebno kako bi se znala
iskoristiti informacija, a najbolje će dostupno znanje iskoristiti mudar čovjek. Jedini način da
podatak dobije smisao i da se otkrije skriveno znanje u golemoj količini podataka je rudarenje
podataka. Učinkovito rudarenje podacima provodi se određenim tehnikama na različitim
vrstama podataka koji su pohranjeni u bazi podataka. Rudarenje podacima je proces
otkrivanja novih korelacija, obrazaca i dubinsko pretraživanje velikih količina podataka
pohranjenih u bazama podataka. Rudarenje podataka tako se ujedno i naziva otkrivanjem
znanja iz baza podataka. CRISP DM model je najčešće korištena metodologija razvoja
procesa rudarenja podacima i definira faze provođenja otkrivanja znanja iz baza podataka,
kao što i definira svaku fazu zadatka i izvedbu svakog zadatka. Algoritmi dubinske analize
dijele u dvije osnovne skupine. Prva skupina su nadgledani algoritmi, odnosno nadgledano
učenje. Metode klasifikacije, odnosno razvrstavanja podataka te metode regresije pripadaju
metodama prve skupine. Drugu skupinu čine nenadgledani algoritmi, drugim riječima
nenadgledano učenje. Metode koje pripadaju drugoj skupini su metode grupiranja, odnosno
klastera te metoda asocijativnih pravila. Na namjernom uzorku od 209 studenta Ekonomskog
fakulteta u Zagrebu provedena je anketa o korištenju društvenih mreža. Rezultati ankete
prikazani su u .arff datoteci čija baza podataka sadrži 20 atributa, od kojih su 3 numerički,
dok je ostalih 17 nominalne vrijednosti. U istraživanju je od ukupno 209 ispitanika, njih 169
ženskog, a 60 muškog spola čija je prosječna starost 21,64 godine. Svi ispitanici koriste
društvenu mrežu Facebook, 131 ispitanik koristi Instagram, 67 njih ima profil na LinkedIn-u,
a samo 15 ispitanika koristi Twitter. Za prikaz rezultata korišteno je stablo odlučivanja.
Smanjivanjem broja atributa dolazi do smanjenja točno klasificiranih podataka. Usporedivši
stablo sa 12 atributa te osnovno stabla sa 20 atributa vidljivo je kako se broj listova smanjio
sa 15 na 10, a ukupna veličina se smanjila sa 23 na 16. Kod stabla s 8 atributa veličina stabla
se u odnosu na početnu veličinu nije promijenila iako je uklonjeno 12 atributa te iznosi 23
dok se broj listova se također smanjio za 2.
Abstract (english) In order to effectively exploit knowledge from large amounts of data helps a wide range of
applications and technologies for data collection and expert analysis known as business
intelligence which is defined as previously concealed knowledge that reveals from operating,
routinely collected business data by using a numerical-logical method and informatical
technology. Large databases are rich in data, but at the same time they are poor in
information. Data is obtained by observation or measurement while providing context to data
obtained information. Knowledge is necessary to know how to take advantage of information
and best available knowledge will ensure wise man. The only way to get a sense to data and
to discover hidden knowledge by a huge amount of data is data mining. Effective data mining
is carried out certain techniques to different types of data that are stored in the database.
Mining data is the process of discovering new correlations, patterns and data mining of large
amounts of data stored in databases. Data mining also is also called the discovery of
knowledge from databases. CRISP DM model is the most widely used methodology
development process of data mining and defines the phases of knowledge discovery from
databases, such as the defining each phase of the task and the performance of each task.
Algorithms depth analysis divided into two basic groups. The first group are supervised
algorithms, or supervised learning. Classification methods and regression methods belong to
the methods of the first group. The second group are unsupervised algorithms, in other words
unsupervised learning. Methods that belong to the second group are the methods of grouping
or clustering methods and association rules. For intentional sample of 209 students at the
Faculty of Economics and businss in Zagreb survey was conducted on the use of social
networks. The survey results are presented in .arff file whose database contains 20 attributes,
of which 3 are numerical, while the other 17 face value. The survey of 209 respondents, 169
female and 60 males with an average age of 21.64 years. All respondents used social network
Facebook, 131 respondents use Instagram, 67 of them have a profile on LinkedIn, and only
15 respondents use Twitter. Model used in this work iz decision tree with different number of
attributes. By reducing the number of attributes number of correcty classified instances is
dropping. Comparing the tree with 12 attributes and basic tree with 20 attributes, it is evident
that the number of sheets decreased from 15 to 10 and the overall size is reduced from 23 to
16. For trees with 8 attributes tree size in relation to the initial size has not changed though is
removed 12 attributes and is 23 while the number of sheets is also reduced from 15 to 13.
Keywords
rudarenje podataka
otkrivanje znanja iz baza podataka
metode dubinske analize
klasifikacija podataka
stablo odlučivanja
društvene mreže
Keywords (english)
data mining
data mining techniques
classification
decision tree
social networks
Language croatian
URN:NBN urn:nbn:hr:148:361385
Study programme Title: Business Economics Study programme type: university Study level: graduate Academic / professional title: magistar/magistra ekonomije (magistar/magistra ekonomije)
Type of resource Text
File origin Born digital
Access conditions Access restricted to students and staff of home institution
Terms of use
Created on 2016-09-30 07:47:12