NO MERGE. Alternative collection map based on distance

[hardingnj]

Hi all,

This is an alternative collection map based purely on geographical distance, ie without any knowledge of political boundaries. Pies are placed at the mean lat/lon of members of that cluster.

Method:

• Using cartopy compute pairwise geographical distance between all samples.
• Build a dendrogram, and cut into clusters using a defined branch length (250km).
• within clusters, compute mean lat/lon and species counts. Plot pies accordingly.

The main differences with the standard map are that:

• The Gambia and Guinea Bissau get grouped.
• Tanzania is split across several other clusters ie with Kenya, Uganda around the lake, and Tanzania only.
• Cameroon is split into several clusters, b/c of geographic distance.

Currently clusters are based on a cut size of 250km. That can be adjusted as needed.

To do:

• Consider land distances and sea distances separately, ie Bioko shouldn't be with some Cameroon samples as it stands now.
• Naming of clusters, though perhaps not required.

Just putting it out there to see if we feel this is potentially worth pursuing. It lacks the simplicity of the per-country map, though it is closer to geographical truth.

[review-notebook-app]

Check out this pull request on 

Review Jupyter notebook visual diffs & provide feedback on notebooks.

---

Powered by ReviewNB