Lab 4

This lab covers an introductory tutorial for Python and, in particular, PySAL. All the content is contained in the IPython notebook Lab-04.ipynb, available from the repository ([raw], [html]) and included in the .zip file with the data.

Data analysis in Python and PySAL

• What is Python?
• Why Python?
  • Flexible
  • Intuitive and high-level
  • Widely used and extended
• How Python?
  • Scripts + interpreter (IPython)
  • IPython notebook
• Basic Python in the notebook
  • Cells (code, markdown…)
  • Libraries
  • Help (inline, called)

Scientific computing in Python

• pandas
  • IO operations
  • Data manipulation
• numpy and scipy
  • Core data structures
  • Statistical functions and low-level methods
• matplotlib
  • Visualization

PySAL Intro

High-level library for advanced spatial analysis

• Shapefile IO
• Spatial weights
  • Create (contiguity and distance)
  • Query
  • Save
  • Combine –> intersection of queen and block weights (example)
• Spatial lag
• Choropleth mapping
• ESDA
  • Global statistics
  • Local statistics

spreg

Replication of all of the models run in Lab 3 with GeoDaSpace:

• Non-spatial models:
  • OLS
  • OLS + White
  • OLS + spatial diagnostics
• Spatial heterogeneity:
  • OLS + spatial fixed effects
  • OLS + spatial regimes
• Spatial dependence:
  • OLS + WX
  • Spatial lag model (IV and ML)
  • Spatial error (GMM and ML)
  • Spatial HAC
• Batch example: spatial diagnostics with several weight matrices

References

• Anselin, L; Amaral, P. V.; Arribas-Bel, D. (2012) Technical Aspects of Implementing GMM Estimation of the Spatial Error Model in PySAL and GeoDaSpace. GeoDa Center Technical Report. [Source code]
• McKinney, W. (2012) Python for Data Analysis. O’Reilly Media.
• McKinney, W. (2013) 10-minute tour of pandas. Vimeo.
• Pandas development team (2014) Pandas documentation. Online resource.