Abstract

The most commonly used statistical models of civil war onset fail to correctly predict most occurrences of this rare event in out-of-sample data. Statistical methods for the analysis of binary data, such as logistic regression, even in their rare event and regularized forms, perform poorly at prediction. We compare the performance of Random Forests with three versions of logistic regression (classic logistic regression, Firth rare events logistic regression, and L1-regularized logistic regression), and find that the algorithmic approach provides significantly more accurate predictions of civil war onset in out-of-sample data than any of the logistic regression models. The article discusses these results and the ways in which algorithmic statistical methods like Random Forests can be useful to more accurately predict rare events in conflict data.

Original languageEnglish (US)
Pages (from-to)87-103
Number of pages17
JournalPolitical Analysis
Volume24
Issue number1
DOIs
StatePublished - Dec 1 2016

Fingerprint

civil war
logistics
regression
event
statistical method
performance

ASJC Scopus subject areas

  • Sociology and Political Science

Cite this

Comparing random forest with logistic regression for predicting class-imbalanced civil war onset data. / Muchlinski, David; Siroky, David; He, Jingrui; Kocher, Matthew.

In: Political Analysis, Vol. 24, No. 1, 01.12.2016, p. 87-103.

Research output: Contribution to journalArticle

@article{6f5aed4e2a144bf191a7224dcd9a25c5,
title = "Comparing random forest with logistic regression for predicting class-imbalanced civil war onset data",
abstract = "The most commonly used statistical models of civil war onset fail to correctly predict most occurrences of this rare event in out-of-sample data. Statistical methods for the analysis of binary data, such as logistic regression, even in their rare event and regularized forms, perform poorly at prediction. We compare the performance of Random Forests with three versions of logistic regression (classic logistic regression, Firth rare events logistic regression, and L1-regularized logistic regression), and find that the algorithmic approach provides significantly more accurate predictions of civil war onset in out-of-sample data than any of the logistic regression models. The article discusses these results and the ways in which algorithmic statistical methods like Random Forests can be useful to more accurately predict rare events in conflict data.",
author = "David Muchlinski and David Siroky and Jingrui He and Matthew Kocher",
year = "2016",
month = "12",
day = "1",
doi = "10.1093/pan/mpv024",
language = "English (US)",
volume = "24",
pages = "87--103",
journal = "Political Analysis",
issn = "1047-1987",
publisher = "Oxford University Press",
number = "1",

}

TY - JOUR

T1 - Comparing random forest with logistic regression for predicting class-imbalanced civil war onset data

AU - Muchlinski, David

AU - Siroky, David

AU - He, Jingrui

AU - Kocher, Matthew

PY - 2016/12/1

Y1 - 2016/12/1

N2 - The most commonly used statistical models of civil war onset fail to correctly predict most occurrences of this rare event in out-of-sample data. Statistical methods for the analysis of binary data, such as logistic regression, even in their rare event and regularized forms, perform poorly at prediction. We compare the performance of Random Forests with three versions of logistic regression (classic logistic regression, Firth rare events logistic regression, and L1-regularized logistic regression), and find that the algorithmic approach provides significantly more accurate predictions of civil war onset in out-of-sample data than any of the logistic regression models. The article discusses these results and the ways in which algorithmic statistical methods like Random Forests can be useful to more accurately predict rare events in conflict data.

AB - The most commonly used statistical models of civil war onset fail to correctly predict most occurrences of this rare event in out-of-sample data. Statistical methods for the analysis of binary data, such as logistic regression, even in their rare event and regularized forms, perform poorly at prediction. We compare the performance of Random Forests with three versions of logistic regression (classic logistic regression, Firth rare events logistic regression, and L1-regularized logistic regression), and find that the algorithmic approach provides significantly more accurate predictions of civil war onset in out-of-sample data than any of the logistic regression models. The article discusses these results and the ways in which algorithmic statistical methods like Random Forests can be useful to more accurately predict rare events in conflict data.

UR - http://www.scopus.com/inward/record.url?scp=84963538255&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84963538255&partnerID=8YFLogxK

U2 - 10.1093/pan/mpv024

DO - 10.1093/pan/mpv024

M3 - Article

AN - SCOPUS:84963538255

VL - 24

SP - 87

EP - 103

JO - Political Analysis

JF - Political Analysis

SN - 1047-1987

IS - 1

ER -