Basics

Summary: this tutorial gives an overview over how to do some of the basic statistical measurements with FSharp.Stats.

Central tendency

A central tendency (or measure of central tendency) is a central or typical value for a probability distribution. It may also be called a center or location of the distribution. Colloquially, measures of central tendency are often called averages.

Mean

For a data set, the arithmetic mean, also called the expected value or average, is the central value of a discrete set of numbers: specifically, the sum of the values divided by the number of values:

\(\bar{x} = \frac{1}{n}\left (\sum_{i=1}^n{x_i}\right ) = \frac{x_1+x_2+\cdots +x_n}{n}\)

mean is available as a Sequence (and other collections) extension, as well as meanBy, which takes an additional converter function:

open FSharp.Stats

let mean1 = 
    [10; 2; 19; 24; 6; 23; 47; 24; 54; 77;]
    |> Seq.meanBy float

28.6

let mean2 = 
    [10.; 2.; 19.; 24.; 6.; 23.; 47.; 24.; 54.; 77.;]
    |> Seq.mean

28.6

Truncated mean

Computes the truncated (trimmed) mean where a given percentage of the highest and lowest values are discarded. In total 2 times the given percentage are discarded:

meanTruncated is available as a Sequence (and other collections) extension, as well as meanTruncatedBy, which takes an additional converter function:

let truncMean1 = 
    [10.; 2.; 19.; 24.; 6.; 23.; 47.; 24.; 54.; 77.;]
    |> Seq.meanTruncated 0.2

24.5

let truncMean2 = 
    [10; 2; 19; 24; 6; 23; 47; 24; 54; 77;]
    |> Seq.meanTruncatedBy float 0.2

34.75

Median

The median is a value separating the higher half from the lower half of a data sample, a population, or a probability distribution. For a data set, it may be thought of as "the middle" value: if you sort the values of a collection by size, the median is the value in central position. Therefore, there are as many bigger values as smaller values than the median in the collection. If there is an even number of observations, then there is no single middle value; the median is then usually defined to be the mean of the two middle values.

median is available as a equence (and other collections) extension:

let median1 = 
    [10; 2; 19; 24; 6; 23; 47; 24; 54; 77;]
    |> Seq.median

23

Harmonic mean

The harmonic mean can be expressed as the reciprocal of the arithmetic mean of the reciprocals of the given set of observations. It is typically appropriate for situations when the average of rates is desired.

\(H = \frac{n}{\frac1{x_1} + \frac1{x_2} + \cdots + \frac1{x_n}} = \frac{n}{\sum\limits_{i=1}^n \frac1{x_i}} = \left(\frac{\sum\limits_{i=1}^n x_i^{-1}}{n}\right)^{-1}.\)

meanHarmonic is available as a sequence (and other collections) extension, as well as meanHarmonicBy, which takes an additional converter function:

let harmonicMean1 = 
    [10.; 2.; 19.; 24.; 6.; 23.; 47.; 24.; 54.; 77.;]
    |> Seq.meanHarmonic

10.01109262

let harmonicMean2 = 
    [10; 2; 19; 24; 6; 23; 47; 24; 54; 77;]
    |> Seq.meanHarmonicBy float

10.01109262

Geometric mean

The geometric mean indicates the central tendency or typical value of a set of numbers by using the product of their values (as opposed to the arithmetic mean which uses their sum). The geometric mean is defined as the nth root of the product of n numbers:

\(\left(\prod_{i=1}^n x_i\right)^\frac{1}{n} = \sqrt[n]{x_1 x_2 \cdots x_n}\)

meanGeometric is available as a sequence (and other collections) extension, as well as meanGeometricBy, which takes an additional converter function:

let geometricMean1 = 
    [10.; 2.; 19.; 24.; 6.; 23.; 47.; 24.; 54.; 77.;]
    |> Seq.meanGeometric

18.9280882

let geometricMean2 = 
    [10; 2; 19; 24; 6; 23; 47; 24; 54; 77;]
    |> Seq.meanGeometricBy float 
 

18.9280882

Dispersion

Dispersion (also called variability, scatter, or spread) is the extent to which a distribution is stretched or squeezed.

Range

The range of a set of data is the difference between the largest and smallest values.

range is available as a sequence (and other collections) extension, as well as rangeBy, which takes an additional converter function:

Note: instead of returning the absolute difference between max and min value, these functions return an interval with these values as boundaries. **

let range1 = 
    [10.; 2.; 19.; 24.; 6.; 23.; 47.; 24.; 54.; 77.;]
    |> Seq.rangeBy float

Closed (2.0, 77.0)

let range2 = 
    [10; 2; 19; 24; 6; 23; 47; 24; 54; 77;]
    |> Seq.rangeBy float

Closed (2.0, 77.0)

Variance and Standard Deviation

The variance

\(s_N^2 = \frac{1}{N} \sum_{i=1}^N \left(x_i - \bar{x}\right)^2\)

and the standard deviation

\(s_N = \sqrt{\frac{1}{N} \sum_{i=1}^N \left(x_i - \bar{x}\right)^2}\)

are measures of dispersion the values of a collection have. While the standard deviation has the same unit as the values of the collection the variance has the squared unit.

varPopulation and stDevPopulation are available as sequence (and other collections) extensions, as well as varPopulationBy and stDevPopulationBy, which take an additional converter function:

let data = [|1.;3.;5.;4.;2.;8.|]

let varPopulation = Seq.varPopulation data

5.138888889

let stdPopulation = Seq.stDevPopulation data

2.266911751

If the full population is not given, the calculation lacks in one degree of freedom, so the Bessel corrected version of the calculation has to be used (results in higher values):

\(s^2 = \frac{1}{N - 1} \sum_{i=1}^N \left(x_i - \bar{x}\right)^2\) for the unbiased variance estimation, and

\(s = \sqrt{\frac{1}{N-1} \sum_{i=1}^N \left(x_i - \bar{x}\right)^2}\) for the corrected standard deviation.

var and stDev are available as sequence (and other collections) extensions, as well as varBy and stDevBy, which take an additional converter function:

let varSample = Seq.var data

6.166666667

let stdSample = Seq.stDev data

2.483277404

Coefficient of variation

The coefficient of variation is the mean-normalized standard deviation:

\(\widehat{c_{\rm v}} = \frac{s}{\bar{x}}\)

It describes the ratio of the standard deviation to the mean. It assists in comparing measurement variability with varying amplitudes. Use only if data is measured with a ratio scale (meaningful zero values and meaningful intervals).

cv is available as a sequence (and other collections) extension, as well as cvBy, which takes an additional converter function:

let sample1 =   [1.;4.;2.;6.;5.;3.;2.;]
let sample2 =   [13.;41.;29.;8.;52.;34.;25.;]

let cvSample1 = Seq.cv sample1

0.5476650327

let cvSample2 = Seq.cv sample2

0.5313890073