Closed pierretoussing closed 4 years ago
In "Related Work" of the PhD Thesis of our former colleague from Institute you can find some interesting state of the art dependency/similarity metrics.
After having found a decent number of different similarity measures, we now have to choose which one's are to keep. This will be done in #21
Most climatologists primarly use Pearson's Correlation or Mutual Information. Those two capture different types of dependencies. While both of them are useful, they also miss certain informations. One goal would be to search and present other similarity metrics for time series, which could be useful for climate data.
The decision which ones are useful and which aren't would be made by the climatologists.
A summary of all the similarity metrics which have been found and implemented can be found here