The project http://cpanratings.perl.org allows us to rate the perl CPAN modules, and it provides interesting information in terms of starts and comments. You can analyze the stats csv file in order to elaborate new stats. The question I'm trying to answer is this: what's the most rated module? ...
Each line of the CSV file give us the name of the module, the rating (a number between 1 and 5) and the reviews counter. I thought about the list order, which field should I use? After a while, I decided to calculate a new value:
RxR = Rating x Reviews
A simple perl script display the list I was asking for (top 30):
# Module Name Rating Reviews RxR
== ============================== ====== ======= ======
1 DBI 4.7 31 145.7
2 Spreadsheet-WriteExcel 4.8 26 124.8
3 Template-Toolkit 4.6 21 96.6
4 Catalyst-Runtime 4.8 20 96
5 Devel-NYTProf 4.9 18 88.2
6 CGI-Application 4.8 18 86.4
7 HTML-Template 4.1 21 86.1
8 WWW-Mechanize 4.6 18 82.8
9 Dancer 3.5 23 80.5
10 XML-Simple 4.4 17 74.8
11 DBD-SQLite 4.6 16 73.6
12 Devel-Cover 4.9 15 73.5
13 Moose 4.9 15 73.5
14 Class-DBI 4.4 16 70.4
15 App-cpanminus 5.0 13 65
16 DateTime 4.6 14 64.4
17 Module-Build 3.2 20 64
18 ack 4.9 13 63.7
19 HTML-Mason 4.9 13 63.7
20 POE 4.9 13 63.7
21 File-Find-Rule 4.5 14 63
22 libwww-perl 4.5 14 63
23 DBD-Oracle 3.7 16 59.2
24 CGI.pm 4.5 13 58.5
25 XML-LibXML 4.5 13 58.5
26 Imager 4.8 12 57.6
27 MIME-Lite 4.8 12 57.6
28 XML-RSS 4.1 14 57.4
29 SQL-Translator 3.9 14 54.6
30 DBIx-Class 4.9 11 53.9
I named it: The top RxR module list.
Do you know how many modules are rated? 2707 (11% of the whole), and 1700 has only one review.
How about the useful comment rate? (The "x out of y found this review helpful"). How about RxRxU where U is the useful percentage average.
ResponderEliminarThis is neat. I'd also like to see a report of negative reviews/ratings of each user, correlated to identify potential trolls and give them a troll icon on the site.
ResponderEliminar