network preprocessing

[martinfleis]

This is what we do now based on the original road networks before doing anything with it

roads = momepy.remove_false_nodes(roads)
roads = roads[~roads.geometry.duplicated()].reset_index()
roads = momepy.remove_false_nodes(roads)

I believe that the Parquet versions of "no node 2" networks should be an output of this, not just a single remove_false_nodes run.

[jGaboardi]

OK, so I will update #80 with:

• refactored remove_degree_2_nodes()
• re-generated data/*/no_degree_2.parquet
• adjusted tests as needed.
• Seems I should remove this chunk from common.continuity(), correct?

Let me know how this sounds

[martinfleis]

Seems I should remove this chunk from common.continuity(), correct?

Yeah, that will be redundant.

[jGaboardi]

We want to drop or keep the index after resetting in line 2?

roads = momepy.remove_false_nodes(roads)
roads = roads[~roads.geometry.duplicated()].reset_index()
roads = momepy.remove_false_nodes(roads)

@martinfleis @anastassiavybornova

I think .reset_index(drop=True) is what we want. Is there a reason to keep the original index for tracking the count dropped? And if so, should that new column be renamed something other than index?

[martinfleis]

Reset. We don't care.

[jGaboardi]

Resolving #80