search

so-wise / weddell_gyre_clusters

Unsupervised classification of Weddell Gyre profiles
MIT License
2 stars 1 forks source link

How sensitive are the results to how the PC model is trained? #119

Closed DaniJonesOcean closed 1 year ago

DaniJonesOcean commented 1 year ago

R1 writes:

...it is not mentioned how large the training data set is, or what the ‘training’ process looks like for the PC model. Are the final PCM conclusions sensitive to how the PC model is trained?

We should expand on this. We could also do a test to see if increasing the number of PCs changes the results. Reducing the number of PCs likely would, but simply because we're not capturing as much of the variability. Let's create a notebook and try this out sometime, perhaps?

DaniJonesOcean commented 1 year ago

Tried this. We increased the number of PCs to 10, and it didn't make any appreciable difference. Capturing 95% of the variability is pretty good