Postupci ansambla stabala odluke

undergraduate thesis

• Title (english): Decision Tree Ensemble Methods
• Author: Darin Dašić
• Mentor: Alan Jović (mentor)
• Committee member: Alan Jović (predsjednik povjerenstva)
• Committee member: Davor Kukolja (član povjerenstva)
• Committee member: Martin Soldić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2017-09-08, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Ansambli stabala odluke su modeli strojnog učenja zasnovani na višestrukim osnovnim modelima, stablima odluke. Postoje dvije glavne metode za izgradnju ansambla: boosting i bagging. Boosting iterativno gradi stabla tako da nove iteracije ovisi o prošloj iteraciji, dok stabla izgrađena bagging metodom ovise o bootstrap uzorku za učenje. Prilikom glasanja konačnog ansambla, boosting provodi težinsko glasanje dok bagging ne. Jedna implementacija bagging metode su slučajne šume, koje dodatno uvode slučajnost prilikom odabira atributa pri izgradnji stabla. U ovom radu opisana je izgradnja slučajne šume, ostvarena je programska implementacija metode te je provedena eksperimentalna usporedba vrednovanja rezultata između ovog ostvarenja i alata Weka na nekoliko skupova podataka.

Abstract (english)

Ensemble methods are based on combining multiple base models. There are two main methods for building an ensemble: boosting and bagging. Base models are constructed iteratively in boosting, where each iteration depends on the performance of the previous one. In bagging, base models are constructed upon bootstrap sample from the training data. When the ensemble makes a decision in boosting each base model casts a weighted vote, whereas in bagging each vote counts the same. Random forests are modification of the bagging method where each CART tree is additionally randomized by selecting a random subspace of the attributes for the split. In this work, random forest algorithm is discussed, an implementation is provided and the results of the implementation are compared to those of WEKA.

• Keywords: slučajne; šume; stablo; odluke; klasifikacija; strojno; učenje; ansambl; boosting; bagging
• Keywords (english): random; forest; decision; tree; classification; machine; learning; ensemble; boosting; bagging
• Language: croatian
• URN:NBN: urn:nbn:hr:168:644422
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2019-03-13 19:54:31