algorithm/JavaScript
1
0
Fork 0
mirror of https://github.com/TheAlgorithms/JavaScript.git synced 2025-07-05 08:16:50 +08:00
Code Issues Packages Projects Releases Wiki Activity

Formatted with standard

Browse Source
This commit is contained in:
MarkSFrancis
2021-05-21 16:47:22 +01:00
parent 55ea4efc1d
commit 4f348ccb5b
1 changed files with 31 additions and 31 deletions
Download Patch File Download Diff File Expand all files Collapse all files

62
Dynamic-Programming/Shuf.js
View File

@ -3,10 +3,10 @@ Given a data set of an unknown size,
Get a random sample in a random order Get a random sample in a random order
It's used in data analytics, often as a way to get a small random sample from a data lake or warehouse, or from a large CSV file It's used in data analytics, often as a way to get a small random sample from a data lake or warehouse, or from a large CSV file
*/ */
function shuf(datasetSource, sampleSize) { function shuf (datasetSource, sampleSize) {
let output = fillBaseSample(datasetSource, sampleSize); const output = fillBaseSample(datasetSource, sampleSize)
return randomizeOutputFromDataset(datasetSource, output); return randomizeOutputFromDataset(datasetSource, output)
} }
/** /**
@ -16,39 +16,39 @@ function shuf(datasetSource, sampleSize) {
* @returns {Array.<T>} The random sample, as an array * @returns {Array.<T>} The random sample, as an array
* @template T * @template T
*/ */
function fillBaseSample(datasetSource, sampleSize) { function fillBaseSample (datasetSource, sampleSize) {
let filledIndexes = []; let filledIndexes = []
let output = new Array(sampleSize); let output = new Array(sampleSize)
// Spread data out filling the array // Spread data out filling the array
while (true) { while (true) {
const iterator = datasetSource.next(); const iterator = datasetSource.next()
if (iterator.done) break; if (iterator.done) break
let insertTo = Math.floor(Math.random() * output.length); let insertTo = Math.floor(Math.random() * output.length)
while (filledIndexes.includes(insertTo)) { while (filledIndexes.includes(insertTo)) {
insertTo++; insertTo++
if (insertTo === output.length) { if (insertTo === output.length) {
insertTo = 0; insertTo = 0
} }
} }
output[insertTo] = { output[insertTo] = {
value: iterator.value, value: iterator.value
}; }
filledIndexes = [...filledIndexes, insertTo]; filledIndexes = [...filledIndexes, insertTo]
if (filledIndexes.length === sampleSize) { if (filledIndexes.length === sampleSize) {
break; break
} }
} }
if (filledIndexes.length < output.length) { if (filledIndexes.length < output.length) {
// Not a large enough dataset to fill the sample - trim empty values // Not a large enough dataset to fill the sample - trim empty values
output = output.filter((_, i) => filledIndexes.includes(i)); output = output.filter((_, i) => filledIndexes.includes(i))
} }
return output.map((o) => o.value); return output.map((o) => o.value)
} }
/** /**
@ -58,22 +58,22 @@ function fillBaseSample(datasetSource, sampleSize) {
* @returns {Array.<T>} The random sample, as an array * @returns {Array.<T>} The random sample, as an array
* @template T * @template T
*/ */
function randomizeOutputFromDataset(datasetSource, output) { function randomizeOutputFromDataset (datasetSource, output) {
const newOutput = [...output]; const newOutput = [...output]
let readSoFar = output.length; let readSoFar = output.length
while (true) { while (true) {
const iterator = datasetSource.next(); const iterator = datasetSource.next()
if (iterator.done) break; if (iterator.done) break
readSoFar++; readSoFar++
const insertTo = Math.floor(Math.random() * readSoFar); const insertTo = Math.floor(Math.random() * readSoFar)
if (insertTo < newOutput.length) { if (insertTo < newOutput.length) {
newOutput[insertTo] = iterator.value; newOutput[insertTo] = iterator.value
} }
} }
return newOutput; return newOutput
} }
const main = () => { const main = () => {
@ -82,15 +82,15 @@ const main = () => {
* @param {number} length The number of data items to generate * @param {number} length The number of data items to generate
* @returns {Iterable<number>} Random iterable data * @returns {Iterable<number>} Random iterable data
*/ */
function* generateRandomData(length) { function * generateRandomData (length) {
const maxValue = Math.pow(2, 31) - 1; const maxValue = Math.pow(2, 31) - 1
for (let i = 0; i < length; i++) { for (let i = 0; i < length; i++) {
yield Math.floor(Math.random() * maxValue); yield Math.floor(Math.random() * maxValue)
} }
} }
const source = generateRandomData(1000); const source = generateRandomData(1000)
const result = shuf(source, 10); const result = shuf(source, 10)
console.log(result) console.log(result)
} }
main() main()