Prediktivna analiza upotrebom metode linearne regresije

master's thesis

• Title (english): Predictive analysis using linear regression
• Author: Karmen Paprić
• Mentor: Marko Banek (mentor)
• Committee member: Marko Banek (predsjednik povjerenstva)
• Committee member: Zoran Skočir (član povjerenstva)
• Committee member: Boris Vrdoljak (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2015-07-03, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Electrical Engineering

Abstract

Da bi dubinska analiza podataka bila što uspješnija potrebno je prvo kvalitetno upoznati i pripremiti podatke. Po potrebi podaci se transformiraju. Prediktivna analiza podataka se koristi za predviđanje vjerojatnih događaja u budućnosti s nekom razinom pouzdanosti. Predviđanje je zapravo samo sugestija onoga što se može dogoditi, ali nikako i garancija da će se baš to doista i dogoditi. Čim je više povijesnih podataka dostupno za izgradnju modela, to će model biti precizniji i kvalitetniji. Na studijskom primjeru regresije prikazani su koraci od upoznavanja s podacima do izvođenje linearne regresije nad konkretnim podatkovnim skupom. Izvedena je transformacija atributa, provjerena korelacija među varijablama te dan grafički prikaz varijabli histogramima. Za provjeru ovisnosti izlazne varijable o ulaznim korišten je t-test te je konačno izvedena linearna regresija nad podacima. Na kraju je automatskom metodom za odabir varijabli za regresiju prikazan konačni model.

Abstract (english)

In order to ensure a successful data mining analysis, the first need is to have a good knowledge of the data and to prepare them for analysis. If it is necessary, data can be transformed. Predictive analysis is used to predict the event in the future with some confidence level. Prediction is actually only a suggestion of what can happen, but not a guarantee that this will really happen. The more historical data is available for the construction of the model, the more precise the resulting model will be. In this thesis we preformed regression on case study data, preforming all steps of data mining analysis, from getting an initial view aimed at understanding the data to creating the final linear regression model on a the data. Attributes were transformed, correlation between variables was tested and graphical presentation of variables with histograms was provided. In order to check the output variables dependence on the input we used t-test, and finally performed a linear regression. The final regression model was constructed using automated method for variable selection.

• Keywords: dubinska analiza podataka; priprema podataka; vrste podataka; transformacija podataka; prediktivna analiza; linearna regresija; korelacija; histogram; koeficijent regresije; odabir varijabli za regresiju
• Keywords (english): data mining; data preparation; data types; data transformation; predictive analysis; linear regression; correlation; histogram; regression coefficient; variable selection for regression
• Language: croatian
• URN:NBN: urn:nbn:hr:168:581385
• Study programme: Title: Information and Communication Technology; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka informacijske i komunikacijske tehnologije (magistar/magistra inženjer/inženjerka informacijske i komunikacijske tehnologije)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2018-02-06 11:04:18