conditional.quantile {verification} | R Documentation |
Conditional Quantile Plot
Description
This function creates a conditional quantile plot as shown in Murphy, et al (1989) and Wilks (1995).
Usage
conditional.quantile(pred, obs, bins = NULL, thrs = c(10, 20), main, ... )
Arguments
pred |
Forecasted value. ([n,1] vector, n = No. of forecasts) |
obs |
Observed value.([n,1] vector, n = No. of observations) |
bins |
Bins for forecast and observed values. A minimum number of values are required to calculate meaningful statistics. So for variables that are continuous, such as temperature, it is frequently convenient to bin these values. ([m,1] vector, m = No. of bins) |
thrs |
The minimum number of values in a bin required to calculate the 25th and 75th quantiles and the 10th and 90th percentiles respectively. The median values will always be displayed. ( [2,1] vector) |
main |
Plot title |
... |
Plotting options. |
Value
This function produces a conditional.quantile plot.
The y
axis represents the observed values, while the x
axis
represents the forecasted values. The histogram along the bottom axis
illustrates the frequency of each forecast.
Note
In the example below, the median line extends beyond the range of
the quartile or 10th and 90th percentile lines. This is because there
are not enough points in each bin to calculate these quartile values.
That is, there are fewer than the limits set in the thrs
input.
Author(s)
Matt Pocernich
References
Murphy, A. H., B. G. Brown and Y. Chen. (1989) Diagnostic Verification of Temperature Forecasts. Weather and Forecasting, 4, 485–501.
Examples
set.seed(10)
m<- seq(10, 25, length = 1000)
frcst <- round(rnorm(1000, mean = m, sd = 2) )
obs<- round(rnorm(1000, mean = m, sd = 2 ))
bins <- seq(0, 30,1)
thrs<- c( 10, 20) # number of obs needed for a statistic to be printed #1,4 quartile, 2,3 quartiles
conditional.quantile(frcst, obs, bins, thrs, main = "Sample Conditional Quantile Plot")
#### Or plots a ``cont.cont'' class object.
obs<- rnorm(100)
pred<- rnorm(100)
baseline <- rnorm(100, sd = 0.5)
A<- verify(obs, pred, baseline = baseline, frcst.type = "cont", obs.type = "cont")
plot(A)