Skip to content

Prediction Confidence - [source code]

Score an ML wind prediction per point: red outside the model's training window, where a feature lies outside the training envelope, or where models disagree by more than 25 %; amber near the window edge, in the envelope's outer 10 %, or where models differ by more than 10 %; green otherwise. The envelope is measured from Dataset Curator CSVs or Verify Prediction logs — none is shipped for the bundled models, because none is published. ML screening prediction, not CFD: use it to compare options, then confirm anything that matters with a Wind Study (planning authorities do not accept surrogate results).

Input

Name Nickname Description Type
Points The points the prediction was made at (absolute Z). Point
Buildings Building geometry, to measure SDF / Bldg_height / Z_relative for the envelope test (as the predictor does). Geometry
Boundary Conditions BC The study's boundary conditions (reference height for U/U_ref). Generic Data
Pedestrian Level PedLevel Pedestrian mounting height [m], as given to the predictor. Number
Envelope Training-envelope sources: Dataset Curator CSVs and/or Verify Prediction logs. Text
Ensemble Optional: the same direction predicted by two or more models, one branch per model, one value per point. Number
Edge Margin Margin Distance [m] from the training window's edge counted as 'near'. Number

Output

Name Nickname Description Type
Level 0 green, 1 amber, 2 red, per point. Integer
Reasons Why each point got its level. Text
Colors Traffic-light colour per point. Colour
Summary Counts per level and what was (and was not) checked. Text