Metrics for climate data

ClimAI / climpy

Tool to manage and climate data for machine learning

BSD 3-Clause "New" or "Revised" License

1 stars 1 forks source link

Metrics for climate data #2

Open ClimAl opened 1 year ago

ClimAl commented 1 year ago

Implement a list of metrics specific to climate data (environmental processes)

[x] Mean Absolute Error
[x] RootMeanSquaredError
[x] Nash–Sutcliffe model efficiency coefficient (NSE)
[ ] Kling-Gupta efficiency (KGE)
[ ] Critical Success Index (CSI) or Threat Score (TS)
[ ] Normalized RMSE
[ ] Bootstrapped ROC
[ ] Spatial Metrics for binary data like Hausdorf Metrics
[ ] Probabilistic Metrics for spatiotemporal data
[ ] All metrics should have a method for the top n percentile

jonathanwider commented 12 months ago

R^2 score (is it the same as NSE?) Ensemble metrics: -> xskillscore

jonathanwider commented 12 months ago

Area weighted versions of the losses?

jonathanwider commented 11 months ago

Anomaly correlation coefficient (ACC)

melioristic commented 11 months ago

@jonathanwider Thanks for your comment! Area weighted version of the losses seems like a nice idea. Yes R2_score and NSE are exactly same in equation but differ in interpretation, we can just call R2_score internally to calculate NSE