A project for NEAS Seminars.
I'll add some real documentation here eventually. For now (taken from the concluding unscientific postscript):
This report was created from a combination of many freely available open-source tools. In the interest of reproducibility, the programs used have been woven into the report itself, ready to be untangled by others on demand.
The source for the report is available as a =.org= file to be read in the Emacs text editor, and in theory the entire report can be automatically produced, from data download through regression analysis to finished =.pdf=, with the use of =GNU Make=.
"make Makefile; make" might work for you. If so, please bear in mind that the code downloads and processes 40+ GB of English Wikipedia data. Best run it on an unused laptop with an external harddrive over a weekend.
In theory. As the sage once said, in theory there is no difference between theory and practice, but in practice there is. The author has made the following effort to explicitly document which programs are necessary to reproduce the analysis with no intervention required. Where appropriate, required version numbers have been included; more common utility programs are assumed to work regardless of version.
-
Primary programs involved were:
- =GNU Emacs=, version 24
- =Python=, version 3
- =SciPy=, =NumPy=, and =matplotlib= packages for =Python=
- \LaTeX with full extensions
-
Common utility programs employed were:
- =make=
- =curl=
- =bzip2=
- =grep=
- =tail=
- =cat=
Have fun.