1#!/sw/bin/gawk -f
2
3BEGIN{
4   args = "";
5   for(i=1; ARGV[i]; i++)
6   {
7      args = args " " ARGV[i];
8   }
9   print "Args: " args;
10}
11
12function success_rate(file    , tmp, succ)
13{
14   pipe = "tsm_classify " args " " file;
15   while ((pipe | getline tmp) > 0)
16   {
17      if(match(tmp, /successes\,/))
18      {
19	 succ = substr(tmp, RSTART+RLENGTH);
20	 print succ;
21      }
22   }
23   close(pipe);
24
25   return 0.0+succ;
26}
27
28BEGIN{
29
30   sum = 0;
31   count = 0;
32   for(i=1; i<=10; i++)
33   {
34      succ[i] = success_rate(sprintf("crossval%02d", i));
35      print succ[i];
36      count++;
37      sum+=succ[i];
38   }
39   avg = sum/count;
40
41   var = 0;
42   for(i=1; i<=10; i++)
43   {
44      var += (succ[i]-avg)*(succ[i]-avg);
45   }
46   var = var/(count-1);
47   dev = sqrt(var);
48
49   print "Success rate: " avg " +/-" dev;
50}
51