src/utils/StatisticUtils.ts

   1 /**
   2  * Computes the average of the given data set.
   3  *
   4  * @param dataSet - Data set.
   5  * @returns The average of the given data set.
   6  * @internal
   7  */
   8 export const average = (dataSet: number[]): number => {
   9   if (Array.isArray(dataSet) && dataSet.length === 0) {
  10     return 0
  11   } else if (Array.isArray(dataSet) && dataSet.length === 1) {
  12     return dataSet[0]
  13   }
  14   return dataSet.reduce((accumulator, nb) => accumulator + nb, 0) / dataSet.length
  15 }
  16
  17 /**
  18  * Computes the median of the given data set.
  19  *
  20  * @param dataSet - Data set.
  21  * @returns The median of the given data set.
  22  * @internal
  23  */
  24 export const median = (dataSet: number[]): number => {
  25   if (Array.isArray(dataSet) && dataSet.length === 0) {
  26     return 0
  27   } else if (Array.isArray(dataSet) && dataSet.length === 1) {
  28     return dataSet[0]
  29   }
  30   const sortedDataSet = dataSet.slice().sort((a, b) => a - b)
  31   return (
  32     (sortedDataSet[(sortedDataSet.length - 1) >> 1] + sortedDataSet[sortedDataSet.length >> 1]) / 2
  33   )
  34 }
  35
  36 // TODO: use order statistics tree https://en.wikipedia.org/wiki/Order_statistic_tree
  37 export const nthPercentile = (dataSet: number[], percentile: number): number => {
  38   if (percentile < 0 && percentile > 100) {
  39     throw new RangeError('Percentile is not between 0 and 100')
  40   }
  41   if (Array.isArray(dataSet) && dataSet.length === 0) {
  42     return 0
  43   }
  44   const sortedDataSet = dataSet.slice().sort((a, b) => a - b)
  45   if (percentile === 0 || sortedDataSet.length === 1) {
  46     return sortedDataSet[0]
  47   }
  48   if (percentile === 100) {
  49     return sortedDataSet[sortedDataSet.length - 1]
  50   }
  51   const percentileIndexBase = (percentile / 100) * (sortedDataSet.length - 1)
  52   const percentileIndexInteger = Math.floor(percentileIndexBase)
  53   // eslint-disable-next-line @typescript-eslint/no-unnecessary-condition
  54   if (sortedDataSet[percentileIndexInteger + 1] != null) {
  55     return (
  56       sortedDataSet[percentileIndexInteger] +
  57       (percentileIndexBase - percentileIndexInteger) *
  58         (sortedDataSet[percentileIndexInteger + 1] - sortedDataSet[percentileIndexInteger])
  59     )
  60   }
  61   return sortedDataSet[percentileIndexInteger]
  62 }
  63
  64 /**
  65  * Computes the sample standard deviation of the given data set.
  66  *
  67  * @param dataSet - Data set.
  68  * @param dataSetAverage - Average of the data set.
  69  * @returns The sample standard deviation of the given data set.
  70  * @see https://en.wikipedia.org/wiki/Unbiased_estimation_of_standard_deviation
  71  * @internal
  72  */
  73 export const stdDeviation = (
  74   dataSet: number[],
  75   dataSetAverage: number = average(dataSet)
  76 ): number => {
  77   if (Array.isArray(dataSet) && dataSet.length === 0) {
  78     return 0
  79   } else if (Array.isArray(dataSet) && dataSet.length === 1) {
  80     return 0
  81   }
  82   return Math.sqrt(
  83     dataSet.reduce((accumulator, num) => accumulator + Math.pow(num - dataSetAverage, 2), 0) /
  84       (dataSet.length - 1)
  85   )
  86 }