@jeremyqzt/nodestats
Version:
Helper methods for stats, combinatorics, permutations, probabilities and matricies
234 lines (211 loc) • 8.56 kB
JavaScript
var zTable = require('./zTable');
/** Class aggregating statistical methods. */
class statsLib{
/**
* Evaluates sum of an array.
* @param {number[]} inArr - The array to sum
* @return {number} Numerical value representing the sum of all array elements
*/
static sum(inArr){
let total = 0;
for (let i = 0; i < inArr.length; i++){
total += inArr[i];
}
return total;
}
/**
* Evaluates geometric sum of an array.
* @param {number[]} inArr - The array to sum
* @return {number} Numerical value representing the geometric sum of all array elements
*/
static geometricSum(inArr){
if (inArr.length === 0){
return 0;
}
let total = inArr[0];
for (let i = 1; i < inArr.length; i++){
total *= inArr[i];
}
return total;
}
/**
* Evaluates statistical mean of an array.
* @param {number[]} inArr - The array to find mean for.
* @return {number} Numerical value representing the mean of all array elements
*/
static mean(inArr){
return statsLib.sum(inArr) / inArr.length;
}
/**
* Evaluates statistical geometric mean of an array.
* @param {number[]} inArr - The array to find geometric mean for.
* @return {number} Numerical value representing the geometric mean of all array elements
*/
static geometricMean(inArr){
let totalGeometric = statsLib.geometricSum(inArr);
return Math.pow(totalGeometric, 1/inArr.length);
}
/**
* Evaluates statistical median of an array.
* @param {number[]} inArr - The array to find median for.
* @return {number} Numerical value representing the median of all array elements
*/
static median(inArr){
let newArr = [...inArr];
let arrSizeEven = newArr.length % 2 === 0;
let midIdx = Math.floor(newArr.length / 2);
newArr.sort(function(a, b) {return a - b;});
return arrSizeEven ? (newArr[midIdx] + newArr[midIdx - 1]) / 2 : newArr[midIdx];
}
/**
* Evaluates statistical mode of an array.
* @param {number[]} inArr - The array to find mode for.
* @return {number} Numerical value representing the mode of all array elements
*/
static mode(inArr){
let retMode = [];
let modeDict = {};
let maxCount = 0;
for (let i = 0; i < inArr.length; i++){
(inArr[i] in modeDict) ? modeDict[inArr[i]]++ : modeDict[inArr[i]] = 1;
if (modeDict[inArr[i]] > maxCount){
retMode = [inArr[i]];
maxCount = modeDict[inArr[i]];
} else if (modeDict[inArr[i]] === maxCount){
retMode.push(inArr[i]);
}
}
return retMode;
}
/**
* Evaluates statistical standard deviation of an array.
* @param {number[]} inArr - The array to find standard deviation for
* @param {{population: boolean}} opt - Option denoting if a population or a sample is provided
* @return {number} Numerical value representing the standard deviation of all array elements
*/
static stdev(inArr, opt = {
"population": true,
}){
let size = inArr.length;
if (!opt["population"]){
size--;
}
let mean = statsLib.mean(inArr);
let total = 0;
for (let i = 0; i < inArr.length; i++){
total += Math.pow((inArr[i] - mean), 2);
}
return Math.sqrt(total / size);
}
/**
* Finds an array element representing the given percentile.
* @param {number} percentile - Percentile to find in the given array (0 <= percentile <= 100)
* @param {number[]} inArr - Array to find the percentile element in
* @return {!number} A numerical value that is a part of the input array representing the given percentile
*/
static absolutePercentile(percentile, inArr){
if (!inArr.length){
return undefined;
}
let newArr = [...inArr];
newArr.sort(function(a, b) {return a - b;});
let idxRank = Math.round((percentile / 100) * newArr.length) - 1;
return newArr[idxRank];
}
/**
* Evaluates the Z score of a given value and a sample or population dataset array.
* @param {number} val - Value to find Z score for
* @param {number[]} inArr - The array representing the population or sample
* @param {{population: boolean}} opt - Option denoting if a population or a sample is provided
* @return {number} Numerical value representing the Z score the given value in the context of the dataset
*/
static zScore(val, inArr, opt = {
"population": true,
}){
let mean = statsLib.mean(inArr);
let stdev = statsLib.stdev(inArr, opt);
return (val - mean) / stdev;
}
/**
* Evaluates the percentile of a given value and a sample or population dataset array.
* @param {number} val - Value to find percentile for
* @param {number[]} inArr - The array representing the population or sample
* @param {{population: boolean}} opt - Option denoting if a population or a sample is provided
* @return {number} Numerical value representing the percentile the given value in the context of the dataset
*/
static percentile(val, inArr, opt = {
"population": true,
}){
let z = statsLib.zScore(val, inArr, opt);
return statsLib.zScorePercentile(z);
}
/**
* Evaluates the percentile of a given value and a sample or population dataset array.
* @param {number} val - Value to find percentile for
* @param {number} mean - The value representing mean of the dataset
* @param {number} stdev - The value representing standard deviation of the dataset
* @return {number} Numerical value representing the percentile the given value in the context of the dataset
*/
static percentileFromMeanAndStdev(val, mean, stdev){
let diff = val - mean;
let zScore = (diff/stdev);
return statsLib.zScorePercentile(zScore);
}
/**
* Evaluates the percentile of a given Z score, maximum is [-3.99, 3.99]. If the value is too large
or too small, it goes to the nearest valid Z score.
* @param {number} score - Z score to find percentile for
* @return {number} Numerical value representing the percentile the given Z score
*/
static zScorePercentile(score){
if (!(typeof score == 'number')){
return undefined;
}
score = (score > 3.99) ? 3.99: score;
score = (score < -3.99) ? -3.99: score;
let combinedIdx = score.toFixed(2);
let idx1 = combinedIdx.substr(0, combinedIdx.length - 1);
let idx2 = combinedIdx.substr(combinedIdx.length - 1, 1);
return zTable.zScoreTab[idx1][idx2];
}
/**
* Evaluates statistical correlation of two arrays.
* @param {number[]} inArr1 - The first array to find correlation for
* @param {number[]} inArr2 - The second array to find correlation for
* @return {number} Numerical value representing the correlation of the 2 arrays
*/
static correlation(inArr1, inArr2){
if (inArr1.length !== inArr2.length){
return 0;
}
const arrSum = (x) => {
let sum = 0;
for (let i = 0; i < x.length; i++){
sum += x[i];
}
return sum;
}
let n = inArr1.length;
let inArr1Sum = arrSum(inArr1);
let inArr2Sum = arrSum(inArr2);
let inArr1Sq = arrSum(inArr1.map(function(num) {
return Math.pow(num, 2)
}));
let inArr2Sq = arrSum(inArr2.map(function(num) {
return Math.pow(num, 2)
}));
let xy = 0;
for (let i = 0; i < n; i++){
xy += inArr1[i] * inArr2[i];
}
let numo = (n*xy) - (inArr1Sum*inArr2Sum);
let deno0 = (n*inArr1Sq) - Math.pow(inArr1Sum,2);
let deno1 = (n*inArr2Sq) - Math.pow(inArr2Sum,2);
let deno = Math.sqrt(deno0*deno1)
let r = numo/deno;
return r;
}
}
module.exports = {
statsLib: statsLib,
};