Labelled Array Development

This is a checklist of things completed, in progress, or to be finished for @ljwolf's GSOC Project, taken from RFC: Clarifying a Core Data Model for PySAL

• pdio: Pandas tools for PySAL, the engine of the labelled array API
  • read/write operations that target Pandas Dataframes
• geocomp: tabular GIS for PySAL
  • conversion interface between pdio's df+geo and geopandas
  • extensions of shapely_ext to tabular GIS operations
  • Wrap/replicate spatial table operations from geopandas (sjoin)
  • NOGR pure python/cython GIS targeting a reduced DE9IM
• work in mainline PySAL
  • make unlabelled array api consistent (shaping, typing)
  • dataframe-capable functions/methods in:
    • weights: from_dataframe, from_iterable for polymorphic constructors
    • spreg: from_formula
    • ESDA: classify_series, exploratory statistics __init__ functions take series/frames
    • core: to_dataframe as a file handle method
  • cg: serialization changes to strandreduce` REQUIRES NOGR does not.
  • close flagged tickets

Before merging current feature branch (Due before midterm deadline):

• Remove unused decorators from pysal.common.
• Add a prefer_shapely or prefer_attributes function to rebuild ops.atomic dispatch on the fly
• Verify that ops.tabular.join has the right docstring & UX
• See if spmin/spmax/sptrace can be removed from sputils:
  • Not worth it. just because they've been made redundant by .min() working on sparse matrix doesn't mean we should make such a basic operation slower for the people who might import from sputils/utils. Though, adding a deprecation warning might be reasonable.
• Return/Readd __author__ to:
  • Contigutiy.py
  • Distance.py
  • weights.py
  • _contW_lists.py
• HAC estimators must have specifically kernel weights, not just any W.
• Pagination on __all__ in pysal.weights.user.
• idVariable must return to rook_from_shapefile
• Superfluous space on l550 of pysal.weights.user