Distributions

Working with statistical distributions

Overview

The Distributions package provides a collection of probability distributions and related functions such as:

  • Sampling from distributions
  • Moments (e.g mean, variance, skewness, and kurtosis), entropy, and other properties
  • Probability density/mass functions (pdf) and their logarithm (logpdf)
  • Moment-generating functions and characteristic functions
  • Maximum likelihood estimation
  • Distribution composition and derived distributions

Getting Started

Load the distributions system with (asdf:load-system :distributions) and generate a sequence of 1000 samples drawn from the standard normal distribution:

(defparameter *rn-samples*
              (nu:generate-sequence '(vector double-float)
			            			 1000
				                     #'distributions:draw-standard-normal))

and plot a histogram of the counts:

(plot:plot
   (vega:defplot normal
       `(:mark :bar
	     :data (:x ,*rn-samples*)
	     :encoding (:x (:bin (:step 0.5)
	                    :field x)
		            :y (:aggregate :count)))))

It looks like there’s an outlier at 5, but basically you can see it’s centered around 0.

To create a parameterised distribution, pass the parameters when you create the distribution object. In the following example we create a distribution with a mean of 2 and variance of 1 and plot it:

(defparameter rn2 (distributions:r-normal 2 1))
(let* ((seq (nu:generate-sequence '(vector double-float) 10000 (lambda () (distributions:draw rn2)))))
  (plot:plot
   (vega:defplot normal-2-1
       `(:mark :bar
	     :data (:x ,seq)
	     :encoding (:x (:bin (:step 0.5)
			            :field x)
		            :y (:aggregate :count))))))

Now that we have the distribution as an object, we can obtain pdf, cdf, mean and other parameters for it:

LS-USER> (mean rn2)
2.0d0
LS-USER> (pdf rn2 1.75)
0.38666811680284924d0
LS-USER> (cdf rn2 1.75)
0.4012936743170763d0

Gamma

In probability theory and statistics, the gamma distribution is a two-parameter family of continuous probability distributions. The exponential distribution, Erlang distribution, and chi-square distribution are special cases of the gamma distribution. There are two different parametrizations in common use:

  • With a shape parameter k and a scale parameter θ.
  • With a shape parameter α = k and an inverse scale parameter β = 1/θ, called a rate parameter.

In each of these forms, both parameters are positive real numbers.

The parametrization with k and θ appears to be more common in econometrics and certain other applied fields, where for example the gamma distribution is frequently used to model waiting times.

The parametrization with α and β is more common in Bayesian statistics, where the gamma distribution is used as a conjugate prior distribution for various types of inverse scale (rate) parameters, such as the λ of an exponential distribution or a Poisson distribution.

When the shape parameter has an integer value, the distribution is the Erlang distribution. Since this can be produced by ensuring that the shape parameter has an integer value > 0, the Erlang distribution is not separately implemented.

PDF

The probability density function parameterized by shape-scale is:

$f(x;k,\theta )={\frac {x^{k-1}e^{-x/\theta }}{\theta ^{k}\Gamma (k)}}\quad {\text{ for }}x>0{\text{ and }}k,\theta >0$,

and by shape-rate:

$f(x;\alpha ,\beta )={\frac {x^{\alpha -1}e^{-\beta x}\beta ^{\alpha }}{\Gamma (\alpha )}}\quad {\text{ for }}x>0\quad \alpha ,\beta >0$

CDF

The cumulative distribution function characterized by shape and scale (k and θ) is:

$F(x;k,\theta )=\int _{0}^{x}f(u;k,\theta ),du={\frac {\gamma \left(k,{\frac {x}{\theta }}\right)}{\Gamma (k)}}$

where $\gamma \left(k,{\frac {x}{\theta }}\right)$ is the lower-incomplete-gamma function.

Characterized by α and β (shape and rate):

$F(x;\alpha ,\beta )=\int _{0}^{x}f(u;\alpha ,\beta ),du={\frac {\gamma (\alpha ,\beta x)}{\Gamma (\alpha )}}$

where $\gamma (\alpha ,\beta x)$ is the lower incomplete gamma function.

Usage

Python and Boost use shape & scale for parameterization. Lisp-Stat and R use shape and rate for the default parametrization. Both forms of parametrization are common. However, since Lisp-Stat’s implementation is based on Boost (because of the restrictive license of R), we perform the conversion $\theta=\frac{1}{\beta}$ internally.

Implementation notes

In the following table k is the shape parameter of the distribution, θ is its scale parameter, x is the random variate, p is the probability and q is (- 1 p). The implementation functions are in the special-functions system.

Function Implementation
PDF (/ (gamma-p-derivative k (/ x θ)) θ)
CDF (incomplete-gamma k (/ x θ))
CDF complement (upper-incomplete-gamma k (/ x θ))
quantile (* θ (inverse-incomplete-gamma k p))
quantile complement (* θ (upper-inverse-incomplete-gamma k p))
mean
variance 2
mode (* (1- k) θ), k>1
skewness (/ 2 (sqrt k))
kurtosis (+ 3 (/ 6 k))
kurtosis excess (/ 6 k)

Example

On average, a train arrives at a station once every 15 minutes (θ=15/60). What is the probability there are 10 trains (occurances of the event) within three hours?

In this example we have:

alpha = 10
theta = 15/60
x = 3

To compute the exact answer:

(distributions:cdf-gamma 3d0 10d0 :scale 15/60)
;=> 0.7576078383294877d0

As an alternative, we can run a simulation, where we draw from the parameterised distribution and then calculate the percentage of values that fall below our threshold, x = 3:

(let* ((rv  (distributions:r-gamma 10 60/15))
       (seq (aops:generate (distributions:generator rv) 10000)))
  (statistics-1:mean (e2<= seq 3))) ;e2<= is the vectorised <= operator
;=> 0.753199999999998d0

Finally, if we want to plot the probability:

(let* ((x    (aops:linspace 0.01d0 10 1000))
       (prob (map 'simple-double-float-vector
		          #'(lambda (x)
		              (distributions:cdf-gamma x 10d0 :scale 15/60))
		          x))
       (interval (map 'vector
		              #'(lambda (x) (if (<= x 3) "0 to 3" "other"))
		              x)))
  (plot:plot
   (vega:defplot gamma-example
       `(:mark :area
	     :data (:x ,x
		        :prob ,prob
		        :interval ,interval)
	      :encoding (:x (:field :x    :type :quantitative :title "Interval (x)")
		             :y (:field :prob :type :quantitative :title "Cum Probability")
		             :color (:field :interval))))))

References

Boost implementation of Gamma
Gamma distribution (Wikipedia)