README.md

In tomhopper/randr: Convenience functions for generating random numbers.

randr

Provides convenience functions for generating random numbers. The main intent is to utilize expert estimations of key distribution characteristics to generate random numbers for monte carlo simulation.

The function names all begin with distribution function from base or VGAM and end with an extension related to the function's operation, as rnorm_within() or runif_digits(). Currently supported are rnorm, runif, rtriangle (via the VGAM package) and rbeta.

*_within() generates within the given range with a confidence_level tolerance interval with lower and upper as the $(\mathrm{confidence_level})/2$ and $(1 - \mathrm{confidence_level})/2$ limits, respectively.

*_between() generates random numbers between the given range, such that $x \in \left[\mathrm{minimum}, \mathrm{maximum}\right]$.

runif_digits generates a vector of length $n$ of uniformly-distributed integers, each with length equal to the given number of digits.

Installation

install.packages("VGAM")
devtools::install_github("tomhopper/numbr")
devtools::install_github("tomhopper/randr")

Examples

Generating normal distributions from max/min estimates

between <- rnorm_between(n = 10000, minimum = 5, maximum = 15)
within <- rnorm_within(n = 10000, lower = 5, upper = 15, confidence_level = 0.8)
within2 <- rnorm_within(n = 10000, lower = 5, upper = 15, confidence_level = 0.99)

hist(between, main = "rnorm_between()", xlab = "estimate")
hist(within, main = "rnorm_within()", xlab = "estimate")
hist(within2, main = "rnorm_within()", xlab = "estimate")

Generating PINs

runif_digits(5, 4)

[1] 4803 3681 5528 7581 1144

Project planning

library(lubridate)
df <- data.frame(a = rbeta_within(100000, 12, 10, 20),
                 b = rbeta_within(100000, 26, 25, 30),
                 c = rbeta_within(100000, 4, 2, 8),
                 d = rbeta_within(100000, 12, 9, 30))
df$total <- apply(df, MARGIN = 1, sum)
mean_total_dur <- mean(df$total)
safe_total_dur <- quantile(df$total, probs = c( 0.95))

delivery_date <- now() + days(as.integer(safe_total_dur))
sums <- apply(X = df, MARGIN = 2, FUN = mean)
sprintf("Planned duration for %s: %.0f days.", c(letters[1:4],"total"), sums)
sprintf("Buffer length: %.0f days", safe_total_dur - mean_total_dur)
sprintf("Delivery date: %s, %.0f days from now.", format(delivery_date, "%Y-%m-%d"), safe_total_dur)

library(tidyr)
library(ggplot2)
df2 <- gather(data = df, key = task, value = duration)
ggplot(df2) +
  geom_histogram(aes(x = duration), binwidth = 1) +
  facet_wrap(~ task, scales = "free")

[1] "Planned duration for a: 14 days."     "Planned duration for b: 27 days."    
[3] "Planned duration for c: 5 days."      "Planned duration for d: 17 days."    
[5] "Planned duration for total: 62 days."
[1] "Buffer length: 14 days"
[1] "Delivery date: 2015-10-20, 75 days from now."

History

1.0.1 : Improved error handling and added unit testing.

1.0.0 : Some function arguments have changed for clarity, breaking code that used the older versions. In particular, the *_between() functions now use minimum and maximum in place of lower and upper. rnorm_between() and runif_between() have been rewritten to produce more random results that fall between minimum and maximum. These functions also produce results that are more consisent with rbeta_between() and rtriangle_between() (VGAM::rtriangle()).

tomhopper/randr documentation built on Nov. 8, 2021, 2:01 p.m.