`scripts/prepare-ml-data.js`

31 lines
import fs from 'fs';
import pool from '../database/connection.js';

async function prepareMLData() {
  const result = await pool.query(`
    SELECT w.reference, w.series, w.year_introduced, ph.year, ph.price, ph.condition
    FROM watches w JOIN price_history ph ON w.id = ph.watch_id
    ORDER BY w.reference, ph.year
  `);

  const csv = ['reference,series,year_introduced,price_year,price,condition'];
  for (const row of result.rows) {
    csv.push(`${row.reference},${row.series},${row.year_introduced},${row.year},${row.price},${row.condition || 'unknown'}`);
  }

  fs.writeFileSync('./data/ml-training-data.csv', csv.join('\n'));
  console.log(`Exported ${result.rows.length} records to ml-training-data.csv`);

  // Split train/test
  const shuffled = result.rows.sort(() => Math.random() - 0.5);
  const splitIdx = Math.floor(shuffled.length * 0.8);
  const train = shuffled.slice(0, splitIdx);
  const test = shuffled.slice(splitIdx);

  fs.writeFileSync('./data/ml-train.json', JSON.stringify(train, null, 2));
  fs.writeFileSync('./data/ml-test.json', JSON.stringify(test, null, 2));
  console.log(`Train: ${train.length}, Test: ${test.length}`);
}

prepareMLData().then(() => process.exit(0)).catch(e => { console.error(e); process.exit(1); });