Verified Pre-Owned Laptops, Desktops & Components48-Hour Escrow Protection | Nationwide Delivery
THE LINKKIỆN VALUATION LAB

Know what shapes the price.

Find a starting asking price from real public used-product listings. Inspect the range, comparable configurations and the held-out model evidence.

The computer you want to price

Device condition (optional)

Use known passport measurements. Leave unknown values blank. Condition changes use simple demo adjustments to the hardware starting price.

The suggestion compares public asking prices and any condition measurements you supply. Your final listing price remains your choice.

A market range, with evidence

Calculate a starting price, suggested asking range and similar public listings. Add passport measurements to see how device condition changes the suggestion.

Evidence behind the estimate

Public advertised used-laptop and complete-desktop asking prices; not completed sale transactions. Snapshot freshness and retailer warranty differ.

These results evaluate the original XGBoost model. The suggested asking range, market comparisons and optional condition adjustments are separate demo guidance.

Held-out asking-price MAE2.387.038 ₫
Held-out percentage error51.5%
Family band coverage94.1%
Median prediction-band width21.845.078 ₫

148 examples · 87 independent groups · 5 source domains.

Evaluation, sources and limitations

Version: xgb-asking-v1-977dd2f3ec9b. Training time: 2026-11-13T13:41:02.130833+00:00.

  • laptopaz.vn: 14
  • laptoptcc.com: 78
  • chotot.com: 45
  • laptop88.vn: 3
  • ducanhpc.com: 8

fit: 83 rows / 43 groups · validation: 20 rows / 10 groups · calibration: 24 rows / 17 groups · test: 21 rows / 17 groups

Median-price baseline MAE: 7.324.286 ₫. Ridge comparison MAE: 2.369.182 ₫. R²: 0.765. Row band coverage: 90.5%. Target family coverage: 90%.

  • Asking prices are not sale outcomes. No time-based test is possible from one collection snapshot.
  • Small, retailer-biased sample; held-out metrics are exploratory and may change substantially.
  • Battery/SSD health, hidden damage, negotiation and warranty adjustments are not learned.
  • Family-group bands require representative exchangeable future families; unknown hardware and extrapolation remain uncertain.
  • Shopping guidance also uses 22 supplemental asking-price snapshots inspected 2026-11-15; these are not extra training rows.
  • Reported evaluation metrics describe the original XGBoost model only. Comparable assistance, device-condition adjustments and the heuristic shopping range have not been independently validated.