src/utils/StatisticUtils.ts

   1 import { mean } from 'rambda'
   2
   3 export const min = (...args: number[]): number =>
   4   args.reduce((minimum, num) => (minimum < num ? minimum : num), Infinity)
   5
   6 export const max = (...args: number[]): number =>
   7   args.reduce((maximum, num) => (maximum > num ? maximum : num), -Infinity)
   8
   9 // TODO: use order statistics tree https://en.wikipedia.org/wiki/Order_statistic_tree
  10 export const nthPercentile = (dataSet: number[], percentile: number): number => {
  11   if (percentile < 0 && percentile > 100) {
  12     throw new RangeError('Percentile is not between 0 and 100')
  13   }
  14   if (Array.isArray(dataSet) && dataSet.length === 0) {
  15     return 0
  16   }
  17   const sortedDataSet = dataSet.slice().sort((a, b) => a - b)
  18   if (percentile === 0 || sortedDataSet.length === 1) {
  19     return sortedDataSet[0]
  20   }
  21   if (percentile === 100) {
  22     return sortedDataSet[sortedDataSet.length - 1]
  23   }
  24   const percentileIndexBase = (percentile / 100) * (sortedDataSet.length - 1)
  25   const percentileIndexInteger = Math.floor(percentileIndexBase)
  26   // eslint-disable-next-line @typescript-eslint/no-unnecessary-condition
  27   if (sortedDataSet[percentileIndexInteger + 1] != null) {
  28     return (
  29       sortedDataSet[percentileIndexInteger] +
  30       (percentileIndexBase - percentileIndexInteger) *
  31         (sortedDataSet[percentileIndexInteger + 1] - sortedDataSet[percentileIndexInteger])
  32     )
  33   }
  34   return sortedDataSet[percentileIndexInteger]
  35 }
  36
  37 /**
  38  * Computes the sample standard deviation of the given data set.
  39  *
  40  * @param dataSet - Data set.
  41  * @param dataSetAverage - Average of the data set.
  42  * @returns The sample standard deviation of the given data set.
  43  * @see https://en.wikipedia.org/wiki/Unbiased_estimation_of_standard_deviation
  44  * @internal
  45  */
  46 export const stdDeviation = (dataSet: number[], dataSetAverage: number = mean(dataSet)): number => {
  47   if (Array.isArray(dataSet) && (dataSet.length === 0 || dataSet.length === 1)) {
  48     return 0
  49   }
  50   return Math.sqrt(
  51     dataSet.reduce((accumulator, num) => accumulator + Math.pow(num - dataSetAverage, 2), 0) /
  52       (dataSet.length - 1)
  53   )
  54 }