Wrapper around <code>pit()</code> for use in data.frames

scoring

Provides a collection of metrics and proper scoring rules
(Tilmann Gneiting & Adrian E Raftery (2007)
<doi:10.1198/016214506000001437>, Jordan, A., Krüger, F., & Lerch, S. (2019)
<doi:10.18637/jss.v090.i12>) within a consistent framework for
evaluation, comparison and visualisation of forecasts.
In addition to proper scoring rules, functions are provided to assess
bias, sharpness and calibration
(Sebastian Funk, Anton Camacho, Adam J. Kucharski, Rachel Lowe, Rosalind
M. Eggo, W. John Edmunds (2019) <doi:10.1371/journal.pcbi.1006785>) of
forecasts.
Several types of predictions (e.g. binary, discrete, continuous) which may
come in different formats (e.g. forecasts represented by predictive samples
or by quantiles of the predictive distribution) can be evaluated.
Scoring metrics can be used either through a convenient data.frame format,
or can be applied as individual functions in a vector / matrix format.
All functionality has been implemented with a focus on performance and is
robustly tested. Find more information about the package in the
accompanying paper (<doi:10.48550/arXiv.2205.07090>).

Nikos Bosse

scoringutils

Utilities for Scoring and Assessing Predictions

Sam Abbott 

Hugo Gruson

Johannes Bracher 

Sebastian Funk

pit function

<dl><dt>data</dt>
<dd>a data.frame with the following columns: <code>true_value</code>,
<code>prediction</code>, <code>sample</code>.</dd>
<dt>by</dt>
<dd>Character vector with the columns according to which the
PIT values shall be grouped. If you e.g. have the columns 'model' and
'location' in the data and want to have a PIT histogram for
every model and location, specify <code>by = c("model", "location")</code>.</dd>
<dt>n_replicates</dt>
<dd>the number of draws for the randomised PIT for
integer predictions.</dd></dl>

Arguments

Probability Integral Transformation (data.frame Format) — pit

<dl>

<dt>data</dt>
<dd>a data.frame with the following columns: <code>true_value</code>,
<code>prediction</code>, <code>sample</code>.</dd>


<dt>by</dt>
<dd>Character vector with the columns according to which the
PIT values shall be grouped. If you e.g. have the columns 'model' and
'location' in the data and want to have a PIT histogram for
every model and location, specify <code>by = c("model", "location")</code>.</dd>


<dt>n_replicates</dt>
<dd>the number of draws for the randomised PIT for
integer predictions.</dd>

</dl>

pit: Probability Integral Transformation (data.frame Format)

Description

Usage

Value

Arguments

Details

References

Examples