Fetching the paper…

Reading the bibliography…

2019

Macro F1 and Macro F1

Opitz, Juri, Burst, Sebastian

Original
PDF
Post on X

Understand

The 'macro F1' metric is frequently used to evaluate binary, multi-class and multi-label classification problems.

Yet, we find that there exist two different formulas to calculate this quantity.
In this note, we show that only under rare circumstances the two computations can be considered equivalent.
More specifically, one formula well 'rewards' classifiers which produce a skewed error type distribution.

Macro F1 and Macro F1 · Around

Built on

A systematic analysis of performance measures for classification tasks
Marina Sokolova and Guy Lapalme · 2009
Earlier work this paper cites.
A comparative analysis of classification methods to multi-label tasks in different application domains
A Santos, A Canuto, and Antonino Feitosa Neto · 2011
Earlier work this paper cites.
Optimal thresholding of classifiers to maximize f1 measure
Zachary C Lipton, Charles Elkan, and Balakrishnan Naryanaswamy · 2014
Earlier work this paper cites.

Similar

Semeval-2015 task 10: Sentiment analysis in twitter
Sara Rosenthal, Preslav Nakov, Svetlana Kiritchenko, Saif Mohammad, Alan Ritter, and Veselin Stoyanov · 2015
Cited alongside, same era.
A unified view of multi-label performance measures
Original
Xi-Zhu Wu and Zhi-Hua Zhou · 2017
Cited alongside, same era.

Then

Neural-davidsonian semantic proto-role labeling
Rachel Rudinger, Adam Teichert, Ryan Culkin, Sheng Zhang, and Benjamin Van Durme · 2018
Later among the works it cites.
An argument-marker model for syntax-agnostic proto-role labeling
Juri Opitz and Anette Frank · 2019
Closest in time.

Beyond the bibliography

alphaXiv searches the wider corpus for related work and actual follow-ups.

Open on alphaXiv

alphaXiv is searching for related work…