Usporedba algoritama za strojno učenje u procesu izgradnje prediktivnih modela za predviđanje odlaska klijenata

master's thesis

• Title (english): Comparison of machine learning algorithms in the process of predictive model creation for client churn prediction
• Author: Aleksandar Tunjić
• Mentor: Boris Vrdoljak (mentor)
• Committee member: Boris Vrdoljak (predsjednik povjerenstva)
• Committee member: Mirjana Domazet-Lošo (član povjerenstva)
• Committee member: Ljiljana Brkić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2016-09-29, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Privlačenje novog korisnika u telekom industriji je vrlo skupo, do nekoliko puta skuplje nego zadržavanje postojećeg. Zbog toga se telekom operateri okreću dubinskoj analizi podataka kako bi se iz podataka o korištenju usluga pokušalo predvidjeti da li će neki korisnik prijeći konkurenciji. Postoji više metoda dubinske analize podataka, a razlikuju se po pristupu u složenosti. U ovom radu je provedena analiza i čišćenje podataka telekom operatera, te su izgrađeni i uspoređeni sljedeći modeli: logistička regresija, PLS, slučajne šume i GBM. Također, provedena je i usporedba navedenih algoritama nad izvedenim podacima dobivenim generatorom varijabli.

Abstract (english)

Attaining a new customer in telecom industry is a few times more expensive than retaining an existing customer. This is why telecom operators use data mining approach to try predict if a customer is going to stop using its services and become a client of a competing telecom. There are few data mining methods and they all differ in approach and complexity. In this thesis the process of data analysis and cleaning was done followed by building models: logistic regression, PLS, random forests and GBM. A comparison of results was done on data attained from original and generated variables.

• Keywords: Dubinska analiza podataka; telekomunikacijska industrija; prediktivni algoritmi
• Keywords (english): Data mining; telecom industry; predictive algorithms; churn
• Language: croatian
• URN:NBN: urn:nbn:hr:168:255243
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2018-02-06 14:11:51