search

EQWorks / locussdk-collab

Community collaboration on ideas and practical workflows around `locussdk` usage
1 stars 0 forks source link

recipes - add raster-vector function, clean logic in graph funcs #9

Closed KrisCowie closed 2 years ago

KrisCowie commented 2 years ago

Added a raster-vector function that will take in the first layer of a raster geotiff, and aggregate the values to a given list of polygons. I.e. given a population geotiff from here for example https://ghsl.jrc.ec.europa.eu/download.php?ds=pop , and a geodataframe which contains the geometries of FSAs, it will aggregate the population and add a column to the gdf with the summed pop for each polygon. Cleaned up some logic in the graphing functions that was breaking in edge cases. Finally, added importlib to the scaling function so users can enter whichever sklearn.preprocessing module they like, and it will still work, without having to account for all of them manually (unfortunately, the same trick doesn't work with the models, as they come from too many different packages).

woozyking commented 2 years ago

also lift the PR up from draft mode when you think it's ready for review

KrisCowie commented 2 years ago

Alright, just did one final commit with a normalization function, so users can select a field to first normalize their data around, and then scale it appropriately - so in our example, we can enter population count, it will divide each of the numeric features by that pop_count, and produce per capita values. Then it continues as normal