# HG changeset patch
# User Arnaud Bergeron <abergeron@gmail.com>
# Date 1273513451 14400
# Node ID 227ebc0be7aeddff33a988824b77238a1bad3545
# Parent  c1df23c98eb69de03073953dad787a410d2e5483
Add a graph for the NIST training set and normalize the values.

diff -r c1df23c98eb6 -r 227ebc0be7ae scripts/stat_graph.py
--- a/scripts/stat_graph.py	Mon May 10 11:41:02 2010 -0400
+++ b/scripts/stat_graph.py	Mon May 10 13:44:11 2010 -0400
@@ -7,11 +7,13 @@
 from ift6266 import datasets
 
 nistp_valid = stats.itemfreq(datasets.PNIST07().valid(10000000).next()[1])
+nistp_valid[:,1] /= sum(nistp_valid[:,1])
 nist_valid = stats.itemfreq(datasets.nist_all().valid(10000000).next()[1])
+nist_valid[:,1] /= sum(nist_valid[:,1])
 nist_test = stats.itemfreq(datasets.nist_all().test(10000000).next()[1])
-print 'nistp_valid', sum(nistp_valid[:,1])
-print 'nist_valid', sum(nist_valid[:,1])
-print 'nist_test', sum(nist_test[:,1])
+nist_test[:,1] /= sum(nist_test[:,1])
+nist_train = stats.itemfreq(datasets.nist_all().train(100000000).next()[1])
+nist_train[:,1] /= sum(nist_train[:,1])
 
 xloc = numpy.arange(62)+0.5
 
@@ -23,10 +25,10 @@
     bar(xloc, data, width=width)
     xticks([])
     for x, l in zip(xloc, labels):
-        text(x+width/2, -250, l, horizontalalignment='center', verticalalignment='baseline')
+        text(x+width/2, -0.004, l, horizontalalignment='center', verticalalignment='baseline')
 #    xticks(xloc+width/2, labels, verticalalignment='bottom')
     xlim(0, xloc[-1]+width*2)
-    ylim(0, 7000)
+    ylim(0, 0.1)
 
     savefig(fname)
 
@@ -34,3 +36,4 @@
 makegraph(nistp_valid[:,1], 'nistpvalidstats.png')
 makegraph(nist_valid[:,1], 'nistvalidstats.png')
 makegraph(nist_test[:,1], 'nistteststats.png')
+makegraph(nist_train[:,1], 'nisttrainstats.png')