evaluate_multiclass

Get performance metrics for multiclass predictions

Aims to make machine learning in healthcare as easy as possible.
You can develop customized, reliable, high-performance machine learning models with minimal code.
Models are created with automatic preprocessing, hyperparameter tuning, and
algorithm selection (between
'xgboost' Chen, T. & Guestrin, C. (2016) <arXiv:1603.02754>,
'ranger' Wright, M. N., & Ziegler, A. (2017) <doi:10.18637/jss.v077.i01>,
and 'glm' Friedman J, Hastie T, Tibshirani R. (2010) <doi:10.18637/jss.v033.i01>)
so that they can be easily
put into production. Additionally, there are tools to help understand how a model makes its
predictions, select prediction threshholds for operational use, and evaluate model performance over time.
Code uses 'tidyverse' syntax and most methods have an associated visualization.

Mike Mastanduno

healthcareai

Tools for Healthcare Machine Learning

Levi Thatcher

Michael Levy

Taylor Larsen

Taylor Miller

Rex Sumsion

evaluate_multiclass function

<dl><dt>predicted</dt>
<dd>Vector of predicted probabilities</dd>
<dt>actual</dt>
<dd>Vector of realized outcomes, must be 0/1</dd></dl>

Arguments

Get performance metrics for multiclass predictions — evaluate_multiclass

<dl>

<dt>predicted</dt>
<dd>Vector of predicted probabilities</dd>


<dt>actual</dt>
<dd>Vector of realized outcomes, must be 0/1</dd>

</dl>

evaluate_multiclass: Get performance metrics for multiclass predictions

Description

Usage

Value

Arguments

Examples