perf_pairwise

Additional performance using appropriate metric for pairwise
distribution fit.

nonparametric

tree

nonlinear

survival

Extensions to Freund and Schapire's AdaBoost algorithm, Y. Freund
and R. Schapire (1997) <doi:10.1006/jcss.1997.1504> and Friedman's gradient
boosting machine, J.H. Friedman (2001) <doi:10.1214/aos/1013203451>.
Includes regression methods for least squares, absolute loss,
t-distribution loss, quantile regression, logistic, Poisson,
Cox proportional hazards partial likelihood, AdaBoost
exponential loss, Huberized hinge loss,
and Learning to Rank measures (LambdaMART).

Ridgeway Greg

gbm3

Generalized Boosted Regression Models

James Hickey

Paul Metcalfe

Greg Ridgeway

Stefan Schroedl

Harry Southworth

Terry Therneau

perf_pairwise function

<dl><dt>y</dt>
<dd>responses used for fit.</dd>
<dt>f</dt>
<dd>the predicted responses.</dd>
<dt>group</dt>
<dd>the indices specifiying which response variables
belong to which groups.</dd>
<dt>metric</dt>
<dd>What type of performance measure to compute in
<code>perf_pairwise</code>. Can take values "ir_measure_conc",
"ir_measure_mrr", "ir_measure_map" or "ir_measure_ndgc".</dd>
<dt>w</dt>
<dd>is the weights of each observation.</dd>
<dt>max_rank</dt>
<dd>the maximum rank considered in the groups measure.</dd></dl>

Arguments

Greg Ridgeway <a href="/link/gregridgeway%40gmail.com?package=gbm3&version=3.0" data-mini-rdoc="gbm3::gregridgeway@gmail.com">gregridgeway@gmail.com</a>

Author

Performance for pairwise — perf_pairwise

<dl>

<dt>y</dt>
<dd>responses used for fit.</dd>


<dt>f</dt>
<dd>the predicted responses.</dd>


<dt>group</dt>
<dd>the indices specifiying which response variables
belong to which groups.</dd>


<dt>metric</dt>
<dd>What type of performance measure to compute in
<code>perf_pairwise</code>. Can take values "ir_measure_conc",
"ir_measure_mrr", "ir_measure_map" or "ir_measure_ndgc".</dd>


<dt>w</dt>
<dd>is the weights of each observation.</dd>


<dt>max_rank</dt>
<dd>the maximum rank considered in the groups measure.</dd>

</dl>

Greg Ridgeway <a href='mailto:gregridgeway@gmail.com'>gregridgeway@gmail.com</a>

perf_pairwise: Performance for pairwise

Description

Usage

Value

Arguments

Author

See Also