get_pit: Probability integral transformation (data.frame version)

In epiforecasts/scoringutils: Utilities for Scoring and Assessing Predictions

Probability integral transformation (data.frame version)

Description

Compute the Probability Integral Transformation (PIT) for validated forecast objects.

Usage

get_pit(forecast, by, n_replicates = 100)

Arguments

forecast	A forecast object (a validated data.table with predicted and observed values, see as_forecast())
by	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 input data and want to have a PIT histogram for every model and location, specify by = c("model", "location").
n_replicates	The number of draws for the randomised PIT for discrete predictions. Will be ignored if forecasts are continuous.

Value

A data.table with PIT values according to the grouping specified in by.

References

Sebastian Funk, Anton Camacho, Adam J. Kucharski, Rachel Lowe, Rosalind M. Eggo, W. John Edmunds (2019) Assessing the performance of real-time epidemic forecasts: A case study of Ebola in the Western Area region of Sierra Leone, 2014-15, \Sexpr[results=rd]{tools:::Rd_expr_doi("10.1371/journal.pcbi.1006785")}

Examples

result <- get_pit(as_forecast(example_sample_continuous), by = "model")
plot_pit(result)

# example with quantile data
result <- get_pit(as_forecast(example_quantile), by = "model")
plot_pit(result)

epiforecasts/scoringutils documentation built on April 23, 2024, 4:56 p.m.