Comments on R-Chart: Google AI Challenge: Languages Used by the Best Pr...

#Some other metrics that take into account the ran...

2011-06-19T23:24:01.826-07:00

#Some other metrics that take into account the rank and number of people using the language

#Median ranks by language (Low numbers indicate that programmers using that language were good.)
> sort(sapply(levels(df$language),function(a) median(df$rank[df$language==a])));plot(.Last.value)
C Scala Javascript Haskell Lisp OCaml PHP
1123.5 1155.0 1196.0 1493.0 1748.0 1799.5 2086.0
Perl Ruby Java C++ C# Python Go
2128.0 2164.0 2226.0 2420.0 2438.0 2520.0 3242.5
Groovy
4250.0
#Chance that a particular entry is in the top 100 assuming that it depends only on language (High numbers indicate that programmers using that language were good.)
> a<-table(df$language[df$rank<=100])/table(df$language);sort(round(a[a>0],2));plot(.Last.value)

C# Haskell Java Python C++ Lisp OCaml C
0.02 0.02 0.02 0.02 0.03 0.03 0.08 0.17

It would also be much more useful to draw the firs...

2010-12-06T05:35:25.492-08:00

It would also be much more useful to draw the first chart in your post as a cumulative. This way it would be much easier to see the distribution of languages across the ranking positions.

I believe, it would be also interesting to see sta...

2010-12-06T04:36:32.305-08:00

I believe, it would be also interesting to see statistics based on the points gained by the contest entrants, not just their ranking positions.

I've made these charts here based on the data you've published (in Excel though), but using the points instead of rankings:

http://0000b.blogspot.com/2010/12/my-google-ai-challenge-participation.html

Of course, it assumes, that points gained are somehow representative of the actual performance of the bot (I believe they are to some extent, at least). One of the charts also demonstrates the inequalities among the points distribution (1st and 2nd places have 300 points of difference, while in the middle the points are distributed much more closely).

Java, C++, then C#. Not surprisingly, the graphs s...

2010-12-05T10:04:07.852-08:00

Java, C++, then C#. Not surprisingly, the graphs say that:
http://wismuth.com/lang/languages.html
http://langpop.com/

The axis scaling for all the charts should probabl...

2010-12-04T10:20:43.454-08:00

The axis scaling for all the charts should probably be the same.

The first graph illustrates a severe weakness with...

2010-12-03T06:55:50.397-08:00

The first graph illustrates a severe weakness with HSV color palettes. It is barely possible to distinguish 6 different colors, but more than that is almost impossible.