1 | import { factory } from '../../utils/factory'
|
2 |
|
3 | const name = 'std'
|
4 | const dependencies = ['typed', 'sqrt', 'variance']
|
5 |
|
6 | export const createStd = /* #__PURE__ */ factory(name, dependencies, ({ typed, sqrt, variance }) => {
|
7 | /**
|
8 | * Compute the standard deviation of a matrix or a list with values.
|
9 | * The standard deviations is defined as the square root of the variance:
|
10 | * `std(A) = sqrt(variance(A))`.
|
11 | * In case of a (multi dimensional) array or matrix, the standard deviation
|
12 | * over all elements will be calculated by default, unless an axis is specified
|
13 | * in which case the standard deviation will be computed along that axis.
|
14 | *
|
15 | * Additionally, it is possible to compute the standard deviation along the rows
|
16 | * or columns of a matrix by specifying the dimension as the second argument.
|
17 | *
|
18 | * Optionally, the type of normalization can be specified as the final
|
19 | * parameter. The parameter `normalization` can be one of the following values:
|
20 | *
|
21 | * - 'unbiased' (default) The sum of squared errors is divided by (n - 1)
|
22 | * - 'uncorrected' The sum of squared errors is divided by n
|
23 | * - 'biased' The sum of squared errors is divided by (n + 1)
|
24 | *
|
25 | *
|
26 | * Syntax:
|
27 | *
|
28 | * math.std(a, b, c, ...)
|
29 | * math.std(A)
|
30 | * math.std(A, normalization)
|
31 | * math.std(A, dimension)
|
32 | * math.std(A, dimension, normalization)
|
33 | *
|
34 | * Examples:
|
35 | *
|
36 | * math.std(2, 4, 6) // returns 2
|
37 | * math.std([2, 4, 6, 8]) // returns 2.581988897471611
|
38 | * math.std([2, 4, 6, 8], 'uncorrected') // returns 2.23606797749979
|
39 | * math.std([2, 4, 6, 8], 'biased') // returns 2
|
40 | *
|
41 | * math.std([[1, 2, 3], [4, 5, 6]]) // returns 1.8708286933869707
|
42 | * math.std([[1, 2, 3], [4, 6, 8]], 0) // returns [2.1213203435596424, 2.8284271247461903, 3.5355339059327378]
|
43 | * math.std([[1, 2, 3], [4, 6, 8]], 1) // returns [1, 2]
|
44 | * math.std([[1, 2, 3], [4, 6, 8]], 1, 'biased') // returns [0.7071067811865476, 1.4142135623730951]
|
45 | *
|
46 | * See also:
|
47 | *
|
48 | * mean, median, max, min, prod, sum, variance
|
49 | *
|
50 | * @param {Array | Matrix} array
|
51 | * A single matrix or or multiple scalar values
|
52 | * @param {string} [normalization='unbiased']
|
53 | * Determines how to normalize the variance.
|
54 | * Choose 'unbiased' (default), 'uncorrected', or 'biased'.
|
55 | * @param dimension {number | BigNumber}
|
56 | * Determines the axis to compute the standard deviation for a matrix
|
57 | * @return {*} The standard deviation
|
58 | */
|
59 | return typed(name, {
|
60 | // std([a, b, c, d, ...])
|
61 | 'Array | Matrix': _std,
|
62 |
|
63 | // std([a, b, c, d, ...], normalization)
|
64 | 'Array | Matrix, string': _std,
|
65 |
|
66 | // std([a, b, c, c, ...], dim)
|
67 | 'Array | Matrix, number | BigNumber': _std,
|
68 |
|
69 | // std([a, b, c, c, ...], dim, normalization)
|
70 | 'Array | Matrix, number | BigNumber, string': _std,
|
71 |
|
72 | // std(a, b, c, d, ...)
|
73 | '...': function (args) {
|
74 | return _std(args)
|
75 | }
|
76 | })
|
77 |
|
78 | function _std (array, normalization) {
|
79 | if (array.length === 0) {
|
80 | throw new SyntaxError('Function std requires one or more parameters (0 provided)')
|
81 | }
|
82 |
|
83 | try {
|
84 | return sqrt(variance.apply(null, arguments))
|
85 | } catch (err) {
|
86 | if (err instanceof TypeError && err.message.indexOf(' variance') !== -1) {
|
87 | throw new TypeError(err.message.replace(' variance', ' std'))
|
88 | } else {
|
89 | throw err
|
90 | }
|
91 | }
|
92 | }
|
93 | })
|