Sector categorization using gradient boosted trees trained on fundamental firm data

Ming Fang; Lilian Kuo; Frank Shih; Stephen Taylor

doi:10.3233/AF-200308

Sector categorization using gradient boosted trees trained on fundamental firm data

Ming Fang, Lilian Kuo, Frank Shih, Stephen Taylor

Research output: Contribution to journal › Article › peer-review

Abstract

We examine to what extent the GICS sector categorization of equity securities may be systematically reconstructed from historical quarterly firm fundamental data using gradient boosted tree classification. Model complexity and performance tradeoffs are examined and relative feature importance is described. Potential extensions are outlined including ideas to improve feature engineering, validating internal consistency and integrating additional data sources to further improve classification accuracy.

Original language	English (US)
Pages (from-to)	91-99
Number of pages	9
Journal	Algorithmic Finance
Volume	8
Issue number	3-4
DOIs	https://doi.org/10.3233/AF-200308
State	Published - 2020

All Science Journal Classification (ASJC) codes

Finance
Computer Vision and Pattern Recognition
Computer Science Applications
Computational Mathematics

Keywords

GICS sector
financial ratios
fundamental data
gradient boosted trees

Access to Document

10.3233/AF-200308

Cite this

@article{2490727ed4f1495685f11584b9bdc267,

title = "Sector categorization using gradient boosted trees trained on fundamental firm data",

abstract = "We examine to what extent the GICS sector categorization of equity securities may be systematically reconstructed from historical quarterly firm fundamental data using gradient boosted tree classification. Model complexity and performance tradeoffs are examined and relative feature importance is described. Potential extensions are outlined including ideas to improve feature engineering, validating internal consistency and integrating additional data sources to further improve classification accuracy.",

keywords = "GICS sector, financial ratios, fundamental data, gradient boosted trees",

author = "Ming Fang and Lilian Kuo and Frank Shih and Stephen Taylor",

year = "2020",

doi = "10.3233/AF-200308",

language = "English (US)",

volume = "8",

pages = "91--99",

journal = "Algorithmic Finance",

issn = "2158-5571",

publisher = "IOS Press",

number = "3-4",

}

TY - JOUR

T1 - Sector categorization using gradient boosted trees trained on fundamental firm data

AU - Fang, Ming

AU - Kuo, Lilian

AU - Shih, Frank

AU - Taylor, Stephen

PY - 2020

Y1 - 2020

N2 - We examine to what extent the GICS sector categorization of equity securities may be systematically reconstructed from historical quarterly firm fundamental data using gradient boosted tree classification. Model complexity and performance tradeoffs are examined and relative feature importance is described. Potential extensions are outlined including ideas to improve feature engineering, validating internal consistency and integrating additional data sources to further improve classification accuracy.

AB - We examine to what extent the GICS sector categorization of equity securities may be systematically reconstructed from historical quarterly firm fundamental data using gradient boosted tree classification. Model complexity and performance tradeoffs are examined and relative feature importance is described. Potential extensions are outlined including ideas to improve feature engineering, validating internal consistency and integrating additional data sources to further improve classification accuracy.

KW - GICS sector

KW - financial ratios

KW - fundamental data

KW - gradient boosted trees

UR - http://www.scopus.com/inward/record.url?scp=85099450981&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85099450981&partnerID=8YFLogxK

U2 - 10.3233/AF-200308

DO - 10.3233/AF-200308

M3 - Article

AN - SCOPUS:85099450981

SN - 2158-5571

VL - 8

SP - 91

EP - 99

JO - Algorithmic Finance

JF - Algorithmic Finance

IS - 3-4

ER -

Sector categorization using gradient boosted trees trained on fundamental firm data

Abstract

All Science Journal Classification (ASJC) codes

Keywords

Access to Document

Other files and links

Fingerprint

Cite this