fuzzy-matching.js

import {ratio} from 'fuzzball';
import data from './data.js';

const currencyThresholds = {
  USD: 50,
  EUR: 40,
  GBP: 35,
  JPY: 5000,
};

function findMatchingReports(data) {
  const expenses = data.expenses;
  const reports = data.reports;

  for (const expense of expenses) {
    let bestMatchReport = null;
    let bestMatchScore = 0;

    for (const report of reports) {
      let bestReportExpenseScore = 0;
      let bestReportExpense = null;

      // Find the most closely matching expense within the report
      for (const reportExpense of report.expenses) {
        let expenseScore = 0;

        // A - 2 points for matching expense type
        if (expense.expenseType === reportExpense.expenseType) {
          expenseScore += 2;
        }

        // B - 0..1 points for matching vendor
        const vendorScore = ratio(expense.vendor, reportExpense.vendor);
        expenseScore += vendorScore / 100; // normalize to 0..1

        // C - 0..1 points for matching amount (exponential distance)
        const amountDistance = Math.abs(expense.amount - reportExpense.amount);
        const currency = expense.currency || 'USD';
        const amountScore = Math.exp(-amountDistance / currencyThresholds[currency]);
        expenseScore += amountScore * (
          currency === reportExpense.currency ? 1 : 0.2
        );

        // D - dampen the score for each additional expense in the report
        expenseScore *= Math.pow(0.95, report.expenses.length);

        // console.log(
        //   expense.expenseType,
        //   report.name,
        //   reportExpense.expenseType,
        //   expenseScore
        // );

        // Update the best matching expense within the report
        if (expenseScore > bestReportExpenseScore) {
          bestReportExpenseScore = expenseScore;
          bestReportExpense = reportExpense;
        }
      }

      // Update the best matching report if the current report has a higher score
      if (bestReportExpenseScore > bestMatchScore) {
        bestMatchScore = bestReportExpenseScore;
        bestMatchReport = report;
      }
    }

    // Result 🚀
    console.log(expense.expenseType, bestMatchReport.name, bestMatchScore);
    // console.log('---');
  }
}

findMatchingReports(data);