master's thesis
An exploratory analysis of data from the ad delivery system

Marinko Miljević (2016)
Josip Juraj Strossmayer University of Osijek
Faculty of Electrical Engineering, Computer Science and Information Technology Osijek
Department of Software Engineering
Chair of Programming Languages and Systems
Metadata
TitleEksplorativna analiza podataka iz sustava za isporuku oglasa
AuthorMarinko Miljević
Mentor(s)Zdravko Krpić (thesis advisor)
Abstract
U ovom radu je izvršena eksplorativna analiza podataka iz sustava za isporuku oglasa. Analizom se pokušalo otkriti utjecaj pojedinih atributa na ciljni atribut CTR. Provedenom analizom došlo se do zaključaka da atributi koji opisuju datum i vrijeme prikaza oglasa (Hour i DateHour), atributi koji opisuju gdje i kako je oglas prikazan (AdBannerType, Categories, Country i DeviceOS) i atribut Impressions više utječu na ciljni atribut. Tim atributima se u radu posvetila veća pažnja u analizi tako da su dodatno filtrirani te kombinirani međusobno kako bi se osim utjecaja svakog pojedinačnog atributa na ciljni atribut provela i analiza utjecaja različitih kombinacija dva atributa na ciljni atribut. Kao rezultat toga zaključeno je da su uspješnije kategorije one koje se odnose na vijesti, obrazovanje, tehnologiju, piće, hranu, zdravlje, računarstvo, zabavu i sport, da su najčešći i najuspješniji tipovi banera Parallax i Interstitial i da se najveći CTR najčešće ostvaruje od 08:00 do 10:00 sati te od 14:00 do 17:00 sati. Na rezultate analize su utjecali nejednaka raspodjela vrijednosti atributa i niski broj prikaza oglasa. Rezultati dobiveni analizom mogu poslužiti za usporedbe s rezultatima budućih sličnih analiza, a također mogu poslužiti osobama koje bolje poznaju način funkcioniranja sustava za isporuku oglasa u unaprjeđenju tog sustava
KeywordsExploratory data analysis Big Dana Data Science RapidMiner Click-through rate
Parallel title (English)An exploratory analysis of data from the ad delivery system
Committee MembersZdravko Krpić (committee chairperson)
Mirko Kohler (committee member)
Josip Balen (committee member)
GranterJosip Juraj Strossmayer University of Osijek
Faculty of Electrical Engineering, Computer Science and Information Technology Osijek
Lower level organizational unitsDepartment of Software Engineering
Chair of Programming Languages and Systems
PlaceOsijek
StateCroatia
Scientific field, discipline, subdisciplineTECHNICAL SCIENCES
Computing
Process Computing
Study programme typeuniversity
Study levelgraduate
Study programmeGraduate University Study Programme in Computer Engineering
Academic title abbreviationmag.ing.comp.
Genremaster's thesis
Language Croatian
Defense date2016-10-11
Parallel abstract (English)
The task of this thesis was to analyze the data from the system for ad delivering using Exploratory data analysis. The analysis attempts to discover the impact of certain attributes on the target attribute CTR. The analysis led to the conclusion that certain attributes such as attributes that describe date and time when the ad was displayed (Hour and DateHour), attributes that describe where and how the ad was displayed (AdBannerType, Categories, Country and DeviceOS) and attribute Impressions have higher influence on the target attribute. The thesis devoted more attention to the analysis of these attributes in a way that they were further filtered and combined with each other to see the effect of different combinations of two attributes on the target attribute. As a result it has been concluded that categories related to news, education, technology, drink, food, health, computing, entertainment and sport are the most successful, that most common and most successful types of banners are Interstitial and Parallax and that the highest CTR is usually achieved from 08:00 a.m. to 10:00 a.m. and from 02:00 p.m. to 05:00 p.m. Unequal distribution of attribute values and low number of ad impressions influenced the results of the analysis. Results obtained from the analysis can be used for comparisons with the results of similar analysis in future, but also can be used by people with better knowledge of how the ad delivery system works to improve that system
Parallel keywords (Croatian)Eksplorativna analiza podataka veliki skup podataka znanost o podacima RapidMiner klikovni postotak.
Resource typetext
Access conditionOpen access
Terms of usehttp://rightsstatements.org/vocab/InC/1.0/
URN:NBNhttps://urn.nsk.hr/urn:nbn:hr:200:413940
CommitterAnka Ovničević