Every recommendation on FashionFactory is calculated deterministically — no editorial opinion, no sponsored bias. Here is how it works.
Monthly heat and humidity values are normalised from NOAA/WMO seasonal normals (30-year averages). Köppen climate classification is used to group cities into 15 climate classes. Temperature is normalised to a 0–1 scale anchored to global extremes (Oymyakon −67°C to Death Valley 56°C). Humidity uses mean relative humidity from WMO surface observations.
Moisture regain (%) follows ISO 6741-1 standard conditions (20°C, 65% RH). Breathability is derived from fabric air-permeability tests (ISO 9237) normalised 0–1. Wrinkle recovery follows AATCC 66 (Wrinkle Recovery Angle), normalised 0–1. Warmth is derived from CLO value (ISO 11092), normalised against down-insulation baseline.
Each fabric receives a composite score for a given context (climate + usage + garment). The formula weights: thermal comfort (×1.6), breathability (context-amplified), moisture management, formality fit, and wrinkle resistance (amplified for formal contexts). Weights are deterministic and published here — not tuned to promote specific products.
If two climate-context combinations produce the same top-3 fabric ranking and similar scores, they share a canonical URL. This ensures recommendations are only published when they are genuinely distinct.
All content is generated deterministically from structured data — no large language models are used per page. The scoring function, not a language model, determines the recommendation.
Climate profiles are updated when NOAA/WMO releases revised normals (typically every 10 years). Textile data is updated when ISO standards are revised. All dates are tracked in the build metadata.