Institutional Repository of Vilnius University
Išskirčių identifikavimo algoritmai regresijos uždaviniams /
Abstract
dc:descriptionIn this paper we compare and analyze which of the most popular outlier detection methods work best in case of accuracy, masking, swamping and speed. Analyzing literature, 6 methods of detecting outliers have been selected: ”Linear regression method”, DBSCAN, Isolation Forests, LOF, Elliptic Envelope and One-Class Support Vector Machine. To analyze these methods, we have generated 198 datasets that vary in general number of data points, percentage of outliers, outlier error coefficient and border value, and number of clusters for data that is clustered. Three types of datasets have been generated: Linear Regression datasets, Clustered datasets and Clustered Linear Regression datasets. After testing and analyzing the resulting data, we have reached the conclusion that the best methods are LOF and DBSCAN. ”Linear Regression method” was found to perform closely in the case of Linear Regression datasets.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2021
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Jankauskas, Juras,
- Contributors dc:contributor
-
- Petkevičius, Linas
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD107116199&prefLang=en_US