Lazega courts and judges

From InterSciWiki

Jump to: navigation, search

==Programs: Data and Permutations==0 I use a p.bat with these comands and >save then copy the output here where relevant Type this at the command line in unix (Macintcsh) to see the options:

python ./analyze.py --help
python ./sandbox/analyze.py --help
Usage: analyze.py [options]
Options:
 -h, --help            show this help message and exit
 -i INPUT_FILE, --input=INPUT_FILE
                       input filename
 -o PAJEK_OUTPUT_FILE, --output=PAJEK_OUTPUT_FILE
                       write graph to pajek output file
 -p, --permute         permute judge periods
 -r RELINKING_OUTPUT_FILE, --relink=RELINKING_OUTPUT_FILE
                       write relinkings to output file
 -s SEED, --seed=SEED  seed to use for random permutation

Contents

[edit] To generate the net file you type:

In Mac, -r for relinkings

cd .\sandbox
python ./analyze.py -i CHAMBRES_238_corfou.csv -o fr_law.net -r

Inside IDLE for PC

analyze.py -i /Python26/CHAMBRES_238_corfou.csv -o french_law.net
analyze.py -i /Python26/CHAMBRES_238_corfou.csv -o /Python26/french_law.net

[edit] To generate the net file with judges permuted for each year with default seed:

python ./analyze.py -i CHAMBRES_238_corfou.csv -o french_law.net_permuted -p

To generate the net file with judges permuted for each year for a desired randomized seed:

python ./analyze.py -i CHAMBRES_238_corfou.csv -o french_law.net_permuted_321 -p -s 321

Inside IDLE

analyze.py -i /Python26/CHAMBRES_238_corfou.csv -o french_law.net_permuted_321 -p -s 321

[edit] Programs: Relinkings

Scott 07:08, 15 April 2009 (PDT) Attached is a new python script which does all the things it does before but now has the -relink flag so it will generate a file that contains all the relinkings. Let me know if there is a better format you would prefer.

Here is an example of how to use it:

python ./analyze.py -i CHAMBRES_238_corfou.csv -o french_law.net1 -p -s 1 -r relink1.txt

python ./analyze.py -i CHAMBRES_238_corfou.csv -o french_law.net1 -p -s 1 -r relink2.txt

Remember to type 'python analyze.py --help' to see all the options.

DW can do this in \Python26 (PC) or the Macbook

[edit] Process of analysis

first wave was to graph the data
second to get statistics on the data
third wave now done: to randomly permute where judges are assigned keeping constant the number of judges per court
4th wave next: to get statistics on the data
5th: compare the actual to the permuted and we find the patterns.

[edit] Instructions for viewing images

Use iExplorer to view these files and install SVG viewer

[edit] Visualizations

Whole graph no relinking on the left

Horizontal Visualization bicomponent

Vertical Visualization bicomponent

Fr1727bico.net is the name of the Pajek file for the bicomponent.

[edit] Accuracy

These data seem perfectly accurate in that if you adjust the node size so they are clickable and right click each circuit node, the number of judges and the specific number of each judge match exactly the entries in the excel spreadsheet (column search in that year for that court number)

[edit] Relinking graphs

Relinking graphs and program zipped - these are the fragments for fragment search in the *.net file.

[edit] Relinking visualizations

Relink in next 647

Relink in two 6675

Relink in three 82198 (dropping last three years) node sizes reduced.

Full relinking in three maybe 720,000+ because of the years with many judges in the same chamber NO IMAGE

The svg viewer is now enabled WikiSysopWikiSysop 16:07, 18 April 2009 (PDT) substitute htm for svg to restore some earlier images

[edit] Actual and Random

1664 actual 30 gens

1749 Randomized 30 gens

1914 Randomized 32 gens

Relinked in 4 13200+

[edit] Next stage software: format for relink file

Relinking: Just click on the unix shell and type:

cd ~/sandbox
python analyze.py -i <raw data file.csv> -o law.net -r <relink filename>
python analyze.py -i CHAMBRES_238_corfou.csv -o law.net -r true.relink
Judge relinking results

From: "scott w" <scottblanc@gmail.com> Date: Wed, April 22, 2009 8:44 pm

total # years, <sequential or non sequential>, judge pair, years they were linked

Every row has a unique judge pair. If it says 'seq' that means all the years were sequential. IF it says 'nonseq' that means there was at least one non-sequential relinking.

Years they were linked use the -> delimiter.

So for example

6,non-seq,"166 & 235",-> 1-> 3-> 5-> 8-> 10-> 13

means judges 166 & 235 were relinked 'non-sequentially' during 6 different periods, in years offset + 1,3,5,8,10,13.

Make sense?

[edit] Summary

pdf (and doc) files