pandas

##Pandas as a Hard Dependency

This is to start a discussion of adding pandas as a hard dependency for PySAL.

Arguments For

• pandas is the defacto library for high-performance, easy to use data structures and data analysis in the Python ecosystem
• we have it as a soft dependency and it works nicely with the pdio contrib module
• we could leverage a lot of functionality rather than rolling our own
• this would likely enhance interoperability with other packages in the pydata stack (statsmodels, geopandas, scikit-learn, others).
• large downstream users of PySAL are moving to conda for Python package management so pandas installation is less of an issue than previously

Arguments Against

• we currently have pandas as a soft dependency so why add a hard dependency?
• more complex packaging