Investigate dataset discrepancies

MetOffice / XBTs_classification

Project for the classification of eXpendable Bathy Thermographs

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

4 stars 2 forks source link

Investigate dataset discrepancies #20

Closed stevehadd closed 4 years ago

stevehadd commented 4 years ago

I have created a notebook to reproduce the plots in the iMeta paper. The plots currently seems to show differences in the dataset that I am working on compared to the iMeta paper, with substantially fewer unlabelled profiles in the data that I have. It is not clear where this discrepancy could come from so needs to be investigated further in consultation with the Ocean Scientists.

stevehadd commented 4 years ago

The dataset discrepancies have been worked out and the previous paper results have been replicated, so closing this issue.