Discovery — does my compset hold up?
Algorithmic peer suggestions for Home2 Lewisville, scored on chain scale, distance, room count, age, RevPAR, and STR-peer correlation. Use this to defend the current compset or swap in a better-fitting peer.
Sourced from compset_discovery_cache
(algorithm v?, computed ?).
Universe: ? hotels in catchment.
Recall@7
—/—
STR peers found in top 7
Recall@14
—/—
STR peers in top 14
Subject fingerprint
What we're matching candidates against.
Current compset
Hotels in compset_members — how the algorithm ranks them.
| Hotel | Algo rank | Score | Fit | Status | Distance | Rooms | TTM RevPAR |
|---|
Suggested peers — not currently in the compset
Top candidates by algorithm score. Click Add to stage as add; Watchlist to track silently; Reject to feed the noise model.
Filter
Score breakdown
How candidate scores are composed (weights, max points). All candidates ranked on the same dial.
Distance
≤0.5 mi → 20 · <3 mi falls off
RevPAR similarity
±15% → 10 · ±35% → 0
RevPAR correlation (STR)
monthly r² over 24 mo → up to 25
Room count
±25% → 10 · ±50% → 0
Chain scale match
exact → 10 · ±1 tier → 5
Product type
extended-stay vs select-service → ±5
Parent brand
same flag-family bonus → up to 10
Noise penalty
renovation, ramp-up, closure → −0 to −15