FINDING THE FACTORS THAT BEST PREDICT HISTORICAL BATTLE OUTCOMES USING RANDOM FORESTS
Loading...
Authors
Cervantes, Steban A.
Subjects
Battles outcomes
history
force ratio
leadership
random forests
summary statistics
terrain
history
force ratio
leadership
random forests
summary statistics
terrain
Advisors
Lucas, Thomas W.
Date of Issue
2024-06
Date
Publisher
Monterey, CA; Naval Postgraduate School
Language
Abstract
Predicting the outcomes of historical battles presents a significant challenge due to the complex and objective interactions of military strength, tactics, and technology. Other subjective factors, such as leadership, morale, surprise, and terrain, can also play critical roles in determining the outcomes of battles. This study builds upon prior research that used classification trees to predict battle winners using a U.S. Army data set of 660 land battles known as CDB90. This thesis applies random forests to the same data. The CDB90 dataset contains over 130 attributes on most of these battles, of which up to 26 are considered in building our statistical models. A total of 15 sub-models are constructed involving different candidate predictor variables and time periods. We find that random forests consistently outperform classification trees at predicting battle outcomes, i.e., they have higher R-squared and area under the receiver operating curve values. The findings indicate that subjective factors, such as leadership and initiative advantage, generally influence battle outcomes more than objective factors, such as force ratio—though the importance of various factors changes over time.
Type
Thesis
Description
Series/Report No
Department
Organization
Identifiers
NPS Report Number
Sponsors
Funder
Format
Citation
Distribution Statement
Distribution Statement A. Approved for public release: Distribution is unlimited.
Rights
This publication is a work of the U.S. Government as defined in Title 17, United States Code, Section 101. Copyright protection is not available for this work in the United States.