An Evaluation of Classification and Outlier Detection Algorithms

Victoria J. Hodge; James Austin

An Evaluation of Classification and Outlier Detection Algorithms

Computer Science

Research output: Working paper › Preprint

Abstract

This paper evaluates algorithms for classification and outlier detection accuracies in temporal data. We focus on algorithms that train and classify rapidly and can be used for systems that need to incorporate new data regularly. Hence, we compare the accuracy of six fast algorithms using a range of well-known time-series datasets. The analyses demonstrate that the choice of algorithm is task and data specific but that we can derive heuristics for choosing. Gradient Boosting Machines are generally best for classification but there is no single winner for outlier detection though Gradient Boosting Machines (again) and Random Forest are better. Hence, we recommend running evaluations of a number of algorithms using our heuristics.

Original language	English
Number of pages	4
Volume	stat.ML
Publication status	Published - 2 May 2018

Keywords

Anomaly Detection
Classification
Algorithms
outlier detection
TIME-SERIES

Access to Document

ArXiv_1805.00811Other version, 83.5 KBLicence: Unspecified

https://arxiv.org/abs/1805.00811

Cite this

@techreport{e169d08247b24de0bd9081d25ec2d285,

title = "An Evaluation of Classification and Outlier Detection Algorithms",

abstract = "This paper evaluates algorithms for classification and outlier detection accuracies in temporal data. We focus on algorithms that train and classify rapidly and can be used for systems that need to incorporate new data regularly. Hence, we compare the accuracy of six fast algorithms using a range of well-known time-series datasets. The analyses demonstrate that the choice of algorithm is task and data specific but that we can derive heuristics for choosing. Gradient Boosting Machines are generally best for classification but there is no single winner for outlier detection though Gradient Boosting Machines (again) and Random Forest are better. Hence, we recommend running evaluations of a number of algorithms using our heuristics.",

keywords = "Anomaly Detection, Classification, Algorithms, outlier detection, TIME-SERIES",

author = "Hodge, {Victoria J.} and James Austin",

year = "2018",

month = may,

day = "2",

language = "English",

volume = "stat.ML",

type = "WorkingPaper",

}

TY - UNPB

T1 - An Evaluation of Classification and Outlier Detection Algorithms

AU - Hodge, Victoria J.

AU - Austin, James

PY - 2018/5/2

Y1 - 2018/5/2

N2 - This paper evaluates algorithms for classification and outlier detection accuracies in temporal data. We focus on algorithms that train and classify rapidly and can be used for systems that need to incorporate new data regularly. Hence, we compare the accuracy of six fast algorithms using a range of well-known time-series datasets. The analyses demonstrate that the choice of algorithm is task and data specific but that we can derive heuristics for choosing. Gradient Boosting Machines are generally best for classification but there is no single winner for outlier detection though Gradient Boosting Machines (again) and Random Forest are better. Hence, we recommend running evaluations of a number of algorithms using our heuristics.

AB - This paper evaluates algorithms for classification and outlier detection accuracies in temporal data. We focus on algorithms that train and classify rapidly and can be used for systems that need to incorporate new data regularly. Hence, we compare the accuracy of six fast algorithms using a range of well-known time-series datasets. The analyses demonstrate that the choice of algorithm is task and data specific but that we can derive heuristics for choosing. Gradient Boosting Machines are generally best for classification but there is no single winner for outlier detection though Gradient Boosting Machines (again) and Random Forest are better. Hence, we recommend running evaluations of a number of algorithms using our heuristics.

KW - Anomaly Detection

KW - Classification

KW - Algorithms

KW - outlier detection

KW - TIME-SERIES

M3 - Preprint

VL - stat.ML

BT - An Evaluation of Classification and Outlier Detection Algorithms

ER -