Uporaba metod strojnega učenja za izboljšanje spletnega vprašalnika

Rodić, Duško

| | SLO | ENG | Cookies and privacy

Bigger font | Smaller font

First page > Show document

Show document

Title:	Uporaba metod strojnega učenja za izboljšanje spletnega vprašalnika
Authors:	ID Rodić, Duško (Author) ID Kljajić Borštnar, Mirjana (Mentor) More about this mentor...
Files:	MAG_Rodic_Dusko_2020.pdf (4,63 MB) MD5: 161C080968440EA372CEB1653281DB5E PID: 20.500.12556/dkum/f3068e53-2e3b-4491-a3b4-db5ea4db74f9
Language:	Slovenian
Work type:	Master's thesis/paper
Typology:	2.09 - Master's Thesis
Organization:	FOV - Faculty of Organizational Sciences in Kranj
Abstract:	Zbiranje podatkov preko spletnega vprašalnika je v današnjem času pravzaprav stalnica, saj gre za hiter in učinkovit način za zajemanje podatkov iz širše populacije. Pogosto pa so vprašalniki predolgi in tudi kompleksne, zato ne zajemamo ciljne populacije in ne dobimo prave slike o raziskovanem stanju. Poseben primer je zajemanje podatkov z namenom ocene primernosti uporabe zalo-zmogljivega računalništva v oblaku za mala in srednje velika podjetja. Ta vprašalnik je namenjen zajemu vhodnih podatkov za večkriterijski model, ki omogoča oceno potenciala, ki v praksi ni v celoti zaživel. Izhajamo iz predpostavke, da je tudi kompleksnost vprašalnika vplivala na šibek odziv respondentov. Iz tega smo razvili raziskovalno vprašanje: »Ali je mogoče vprašalnik skrajšati s pomočjo strojnega učenja?«. Cilj magistrske naloge je, da s pomočjo metod strojnega učenja skušamo ugotoviti katera vprašanja največ prispevajo k oceni potenciala ter na ta način skrajšati vprašalnik. Problem smo reševali z uporabo metod strojnega učenja. V ta name smo analizirali večkriterijski model, vprašalnik za zajemanje podatkov, odgovore respondentov ter v programu Orange , ki poleg metod strojnega učenja vsebuje tudi vizualizacijo podatkov, IV analizirali prispevek posameznega vprašanja h končni oceni. Rezultati kažejo, da imajo nekateri kriteriji večji vpliv na končno oceno potenciala uporabe zelo-zmogljivega računalništva v oblaku, vendar se ti kriteriji nanašajo na splošne atribute primerov (npr. država, vrsta organizacije). Ob izločitvi trivialnih kriterijev napoved modela še vedno ni dovolj natančna, zato je edini zaključek, ki ga lahko potegnemo, da na pričujočih podatkih ni bilo možno izbrati takega nabora kriterijev oziroma vprašanj, s katerimi bi lahko natančno ocenili potencial uporabe zelo-zmogljivega računalništva v oblaku.
Keywords:	- zelo-zmogljivo računalništvo, - oblak, - strojno učenje, - podatkovno rudarjenje, - Orange (program).
Place of publishing:	Maribor
Year of publishing:	2020
PID:	20.500.12556/DKUM-78228
COBISS.SI-ID:	44128515
NUK URN:	URN:SI:UM:DK:IUEI5N4M
Publication date in DKUM:	23.12.2020
Views:	1268
Downloads:	98
Metadata:
Categories:	FOV
:	Copy citation

Average score:	(0 votes)
Your score:	Voting is allowed only for logged in users.
Share:

Hover the mouse pointer over a document title to show the abstract or click on the title to get all document metadata.

Licences

License:	CC BY-NC-ND 4.0, Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International

Link:	http://creativecommons.org/licenses/by-nc-nd/4.0/
Description:	The most restrictive Creative Commons license. This only allows people to download and share the work for no commercial gain and for no other purposes.
Licensing start date:	20.11.2020

Secondary language

Language:	English
Title:	Use of machine learning techniques for improvement of online questionnaire
Abstract:	Collecting data through an online questionnaire is an everyday task these days, as it is a fast and efficient way to capture data from the general population. However, the questionnaires are often too long and complex, so they do not cover the target population and do not get a true picture of the research situation. Such an example is data capture to assess the suitability of using high-performance cloud computing for small and medium-sized businesses. This questionnaire is intended to capture input data for a multi-criteria model that allows the assessment of a potential that has not been fully implemented in practice. We come from the assumption that the complexity of the questionnaire also influenced the weak response of the respondents. Because of that, we posed following research question: "Is it possible to shorten the questionnaire with the help of machine learning?". The aim of this master's thesis is to use machine learning methods to find out which questions contribute the most to the assessment of potential and thus shorten the questionnaire. To solve a problem we used the machine learning methods. For this purpose, we analyzed a multi-criteria model, a questionnaire for data acquisition, respondents VI answers and in the Orange program, which in addition to machine learning methods also includes data visualization, we analyzed the contribution of each question to the final assessment. The results show that some criteria have a greater impact on the final assessment of the potential use of high-performance computing in cloud, but these criteria relate to the general attributes of the cases (e.g., country, organization size). Excluding general criteria, the model prediction is still not accurate enough, so the only conclusion we can draw is that it was not possible to select such a set of criteria or questions from the presented data that could accurately assess the potential use of high-performance computing in cloud.
Keywords:	- HPC, - Cloud, - machine learning, - data mining, - Orange (program).

Comments

Leave comment

You must log in to leave a comment.

Comments (0)

0 - 0 / 0

There are no comments!

Back