JavaScript中寻找JSON数组最小平均差匹配项的工具库咨询

Great question! You're trying to locate the array in your JSON dataset that has the smallest average absolute difference from a test array. While you can absolutely build a vanilla JavaScript solution for this, there are several libraries that can streamline the process—especially if you might need more advanced similarity metrics down the line. Here are your top options:

1. Lodash #

Lodash is a ubiquitous utility library in JS projects, and while it doesn't have a direct "average difference" function, its helper methods make assembling the logic clean and concise.

Example Usage: #

const _ = require('lodash');

// Your test array (convert strings to numbers first)
const testArr = [73.0, 41.0, 98.0, 43.0];
// Sample JSON dataset
const data = {
  a: ["73.0", "41.0", "98.0", "43.0"],
  b: ["70.0", "45.0", "95.0", "40.0"],
  c: ["80.0", "35.0", "100.0", "45.0"]
};

// Function to calculate average absolute difference between two arrays
const calculateAvgDiff = (arr1, arr2) => {
  const numArr2 = arr2.map(Number);
  const differences = _.zipWith(arr1, numArr2, (x, y) => Math.abs(x - y));
  return _.mean(differences);
};

// Find the key with the smallest average difference
const bestMatch = _.minBy(Object.entries(data), ([key, arr]) => calculateAvgDiff(testArr, arr));

console.log(`Best match: ${bestMatch[0]} with array ${bestMatch[1]}`);

2. Math.js #

Math.js is a dedicated library for mathematical operations, including statistical functions. It simplifies numeric conversions and calculations—perfect for your use case since your dataset arrays contain stringified numbers.

Example Usage: #

const math = require('mathjs');

const testArr = [73.0, 41.0, 98.0, 43.0];
const data = {
  a: ["73.0", "41.0", "98.0", "43.0"],
  b: ["70.0", "45.0", "95.0", "40.0"],
  c: ["80.0", "35.0", "100.0", "45.0"]
};

const calculateAvgDiff = (arr1, arr2) => {
  // Convert string array to numbers and compute element-wise absolute differences
  const numArr2 = math.map(arr2, Number);
  const diffs = math.abs(math.subtract(arr1, numArr2));
  return math.mean(diffs);
};

let minDiff = Infinity;
let bestMatch = null;

for (const [key, arr] of Object.entries(data)) {
  const currentDiff = calculateAvgDiff(testArr, arr);
  if (currentDiff < minDiff) {
    minDiff = currentDiff;
    bestMatch = { key, arr };
  }
}

console.log(`Best match: ${bestMatch.key} with array ${bestMatch.arr}`);

3. TensorFlow.js #

If you're working with very large datasets or might want to expand to more advanced similarity metrics (like cosine similarity) later, TensorFlow.js is a great choice. It handles vector operations efficiently, leveraging WebGL for performance if available.

Example Usage: #

const tf = require('@tensorflow/tfjs');

const testArr = [73.0, 41.0, 98.0, 43.0];
const data = {
  a: ["73.0", "41.0", "98.0", "43.0"],
  b: ["70.0", "45.0", "95.0", "40.0"],
  c: ["80.0", "35.0", "100.0", "45.0"]
};

// Convert test array to a tensor
const testTensor = tf.tensor1d(testArr);

let minDiff = Infinity;
let bestMatch = null;

// Iterate through dataset
for (const [key, arr] of Object.entries(data)) {
  const numArr = arr.map(Number);
  const dataTensor = tf.tensor1d(numArr);
  
  // Calculate element-wise absolute difference, then average
  const diffMean = tf.mean(tf.abs(tf.sub(testTensor, dataTensor))).dataSync()[0];
  
  if (diffMean < minDiff) {
    minDiff = diffMean;
    bestMatch = { key, arr };
  }
  
  dataTensor.dispose(); // Clean up tensor to free memory
}

testTensor.dispose();

console.log(`Best match: ${bestMatch.key} with array ${bestMatch.arr}`);

Bonus: Vanilla JavaScript (No Libraries) #

If you want to avoid adding dependencies, a simple vanilla JS implementation works perfectly for this use case:

const testArr = [73.0, 41.0, 98.0, 43.0];
const data = {
  a: ["73.0", "41.0", "98.0", "43.0"],
  b: ["70.0", "45.0", "95.0", "40.0"],
  c: ["80.0", "35.0", "100.0", "45.0"]
};

const calculateAvgDiff = (arr1, arr2) => {
  let totalDiff = 0;
  for (let i = 0; i < arr1.length; i++) {
    totalDiff += Math.abs(arr1[i] - Number(arr2[i]));
  }
  return totalDiff / arr1.length;
};

let minDiff = Infinity;
let bestMatchKey = '';
let bestMatchArr = [];

for (const [key, arr] of Object.entries(data)) {
  const currentDiff = calculateAvgDiff(testArr, arr);
  if (currentDiff < minDiff) {
    minDiff = currentDiff;
    bestMatchKey = key;
    bestMatchArr = arr;
  }
}

console.log(`Best match: ${bestMatchKey} with array ${bestMatchArr}`);

All these approaches will correctly identify the array with the smallest average difference, just like your example's "a" entry.

内容的提问来源于stack exchange，提问作者Sunny