R/utils.R
binomProbs.RdThis is meant to perform in the same way as quantile() so it can
be a drop in replacement for code using quantile() but using
distributional assumptions.
binomProbs(x, ...)
# Default S3 method
binomProbs(
x,
probs = c(0.025, 0.05, 0.5, 0.95, 0.975),
na.rm = FALSE,
names = TRUE,
onlyProbs = TRUE,
n = 0L,
m = 0L,
pred = FALSE,
piMethod = c("lim"),
M = 5e+05,
tol = .Machine$double.eps^0.25,
ciMethod = c("wilson", "wilsonCorrect", "agrestiCoull", "wald", "wc", "ac"),
...
)numeric vector whose mean and probability based confidence
values are wanted, NA and NaN values are not allowed in numeric
vectors unless na.rm is TRUE.
Arguments passed to default method, allows many different methods to be applied.
numeric vector of probabilities with values in the interval 0 to 1, inclusive. When 0, it represents the maximum observed, when 1, it represents the maximum observed. When 0.5 it represents the expected probability (mean).
logical; if true, any NA and NaN's are removed from
x before the quantiles are computed.
logical; if true, the result has a names attribute.
logical; if true, only return the probability based confidence interval/prediction interval estimates, otherwise return extra statistics.
integer/integerish; this is the n used to calculate the
prediction or confidence interval. When n=0 (default) use the
number of non-NA observations. When calculating the prediction
interval, this represents the number of observations used in the
input ("true") distribution.
integer. When using the prediction interval this represents the number of samples that will be observed in the future for the prediction interval.
Use a prediction interval instead of a confidence
interval. By default this is FALSE.
gives the prediction interval method (currently only lim) from Lu 2020
number of simulations to run for the LIM PI.
tolerance of root finding in the LIM prediction interval
gives the method for calculating the confidence interval.
Can be:
"argestiCoull" or "ac" – Agresti-Coull method. For a 95\ interval, this method does not use the concept of "adding 2 successes and 2 failures," but rather uses the formulas explicitly described in the following link:
https://en.wikipedia.org/wiki/Binomial_proportion_confidence_interval#Agresti-Coull_Interval.
"wilson" – Wilson Method
"wilsonCorrect" or "wc" – Wilson method with continuity correction
"wald" – Wald confidence interval or standard z approximation.
By default the return has the probabilities as names (if
named) with the points where the expected distribution are
located given the sampling mean and standard deviation. If
onlyProbs=FALSE then it would prepend mean, variance, standard
deviation, minimum, maximum and number of non-NA observations.
It is used for confidence intervals with rxode2 solved objects using
confint(mean="binom")
Newcombe, R. G. (1998). "Two-sided confidence intervals for the single proportion: comparison of seven methods". Statistics in Medicine. 17 (8): 857–872. doi:10.1002/(SICI)1097-0258(19980430)17:8<857::AID-SIM777>3.0.CO;2-E. PMID 9595616.
Hezhi Lu, Hua Jin, A new prediction interval for binomial random variable based on inferential models, Journal of Statistical Planning and Inference, Volume 205, 2020, Pages 156-174, ISSN 0378-3758, https://doi.org/10.1016/j.jspi.2019.07.001.
x<- rbinom(7001, p=0.375, size=1)
binomProbs(x)
#> 2.5% 5% 50% 95% 97.5%
#> 0.3723355 0.3741468 0.3836595 0.3932621 0.3951111
# you can also use the prediction interval
# \donttest{
binomProbs(x, pred=TRUE)
#> 2.5% 5% 50% 95% 97.5%
#> 0.3675189 0.3702328 0.3836595 0.3972290 0.3998000
# }
# Can get some extra statistics if you request onlyProbs=FALSE
binomProbs(x, onlyProbs=FALSE)
#> mean var sd n 2.5% 5%
#> 0.3836595 0.2364649 0.4862765 7001.0000000 0.3723355 0.3741468
#> 50% 95% 97.5%
#> 0.3836595 0.3932621 0.3951111
x[2] <- NA_real_
binomProbs(x, onlyProbs=FALSE)
#> mean var sd n 2.5% 5% 50% 95% 97.5%
#> NA NA NA NA NA NA NA NA NA
binomProbs(x, na.rm=TRUE)
#> 2.5% 5% 50% 95% 97.5%
#> 0.3723892 0.3742006 0.3837143 0.3933178 0.3951670