I am working on validating a collection of images from a time series in Google Earth Engine and the confusion matrix that results from validation of the trained image v.s. the landcover classes defined in the validation set is [[0]]. I can't seem to determine what the source of this value is but have hypothesized the following:
1.) The validation data region should not intersect with the trained region
2.) The training data should not intersect with the validation region
Based on inspection in the Earth Engine code editor, neither one of these hypotheses seem to be correct because there are classified regions that violate both hypotheses and still produce a populated confusion matrix.
If anyone has an understanding of what causes an error (confusion) matrix in Earth Engine to return empty that would be awesome. Thanks!
Dear Ian,
I am not sure exactly why you are getting this, but I hope these links will be helpful to understand your problem:
https://gis.stackexchange.com/questions/274166/exporting-classification-error-matrix-google-earth-engine
https://stackoverflow.com/questions/57855813/gee-python-api-export-table-fails-with-unexpected-error
Also, this is our recent publication on landuse heterogeneity. I hope it will be useful to you:
Article Hydrological Response to Agricultural Land Use Heterogeneity...
Cheers
Hi Nikul,
Thank you for those links, unfortunately, I've already checked those issues out and they don't seem to be relevant. I will read your paper when I get the chance.
Thanks!
Dear Ian Dershem,
I also do agrre with the views of Vani Chaturvedi, Ian Dershem and Nikul Kumari and I will appreciate it very much if you can comment so that we will be able to learn the truth.
Best Regards
Desalegn Abraha
- Badges
- Science topic
- Similar topics
- Counseling Psychology
- Confusion