[Jboost-cvs] jboost/scripts error.py,1.4,1.5

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 422-6466

Update of /cvsroot/jboost/jboost/scripts
In directory sc8-pr-cvs6.sourceforge.net:/tmp/cvs-serv11316

Modified Files:
	error.py 
Log Message:
Can now show separate error for positive and negative examples

Index: error.py
===================================================================
RCS file: /cvsroot/jboost/jboost/scripts/error.py,v
retrieving revision 1.4
retrieving revision 1.5
diff -C2 -d -r1.4 -r1.5
*** error.py	23 Oct 2007 22:50:42 -0000	1.4
--- error.py	12 Dec 2007 23:50:29 -0000	1.5
***************
*** 12,36 ****
  	print '\t--bound             show the bound on training error'
  	print '\t--separate          separate the positive and negative examples'
! 	print '\t--scores=score_file the score file (only needed if --separate used)'
! 	print '\t--margins=marg_file the margin file (only needed if --separate used)'

! def do_separate(score_filename, margin_filename):
! 	print 'Parsing scores file'
! 	f = open(score_filename,'r')
! 	lines = f.readlines()
! 	f.close()
! 	
! 	read_iter = False
! 	data = []
! 	for line in lines:
! 		line = line.strip()
! 		if (line[:8] =='End time'):
! 			break
! 		if (read_iter):
! 			data.append(get_score(line))
! 			

!     

--- 12,88 ----
  	print '\t--bound             show the bound on training error'
  	print '\t--separate          separate the positive and negative examples'
! 	print '\t--boost-info=file  The boosting.info file for a data set'

! SEPARATOR = ':'

! def get_margin(line):
! 	m = line.split(SEPARATOR)[1]
! 	m.replace(']','')
! 	m.replace(';','')
! 	m.replace(SEPARATOR,'')
! 	m.replace(' ','')
! 	m.replace('\t','')
! 	return float(m)
! 
! def get_score(line):
! 	m = line.split(SEPARATOR)[2]
! 	m.replace(']','')
! 	m.replace(';','')
! 	m.replace(SEPARATOR,'')
! 	m.replace(' ','')
! 	m.replace('\t','')
! 	return float(m)

+ def getErrorsSingleIter(data):
+ 	tp = 0
+ 	fp = 0
+ 	tn = 0
+ 	fn = 0
+ 
+ 	scores = map(get_score, data)
+ 	margins = map(get_margin, data)
+ 	for score, margin in zip(scores, margins):		
+ 		if score < 0 and margin < 0:
+ 			fn += 1
+ 		if score < 0 and margin > 0:
+ 			tn += 1
+ 		if score > 0 and margin < 0:
+ 			fp += 1
+ 		if score > 0 and margin > 0:
+ 			tp += 1
+ 	return (tp,fp,tn,fn)					
+ 
+ def getErrors(boostfilename):
+     f= open(boostfilename,'r')
+     data= f.readlines()
+     f.close()
+     score_elements = int((string.split(data[0],SEPARATOR))[1].split('=')[1])
+     num_iterations = int((string.split(data[-score_elements-1],SEPARATOR))[0].split('=')[1])
+     print num_iterations
+ 
+     errors = []
+     for iter in range(num_iterations+1):
+ 	    tp,fp,tn,fn = getErrorsSingleIter([x for x in data[iter*(score_elements+1)+1:(iter+1)*(score_elements+1)]])
+ 	    print iter,tp,fp,tn,fn
+ 	    ret = {}
+ 	    if tp + fn > 0:
+ 		    ret['recall'] = ret['sensitivity'] = float(tp) / (tp + fn)
+ 	    else:
+ 		    ret['recall'] = ret['sensitivity'] = 0
+ 	    if tp + fp > 0:
+ 		    ret['precision'] = float(tp) / (tp + fp)
+ 	    else:
+ 		    ret['precision'] = 0
+ 	    if fp + tn > 0:		    
+ 		    ret['fpr'] = float(fp) / (fp + tn)
+ 	    else:
+ 		    ret['fpr'] = 0
+ 	    ret['specificity'] = 1 - ret['fpr']
+ 	    ret['neg_err'] = ret['fpr']
+ 	    ret['pos_err'] = 1 - ret['sensitivity']
+ 	    ret['err'] = float(fp+fn) / (fp+tp+fn+tn)
+ 	    ret['iter'] = iter
+ 	    errors.append(ret)
+     return errors

***************
*** 39,53 ****

      try:
! 	    opts, args= getopt.getopt(sys.argv[1:], '', ['info=','logaxis','bound','separate','scores=','margins='])
      except getopt.GetoptError:
  	    usage()
  	    sys.exit(2)

!     info_filename = logaxis = showbound = show_separate = margin_filename = score_filename = None
      for opt,arg in opts:
  	    if (opt == '--info'):
  		    info_filename = arg
! 	    if (opt == '--margins'):
! 		    margin_filename = arg
  	    if (opt == '--info'):
  		    score_filename = arg
--- 91,105 ----

      try:
! 	    opts, args= getopt.getopt(sys.argv[1:], '', ['info=','logaxis','bound','separate','boost-info='])
      except getopt.GetoptError:
  	    usage()
  	    sys.exit(2)

!     info_filename = logaxis = showbound = show_separate = boost_filename = None
      for opt,arg in opts:
  	    if (opt == '--info'):
  		    info_filename = arg
! 	    if (opt == '--boost-info'):
! 		    boost_filename = arg
  	    if (opt == '--info'):
  		    score_filename = arg
***************
*** 60,69 ****

!     if (show_separate and (margin_filename==None or score_filename==None)):
  	    print 'ERROR: Need to specify score and margin file if showing positive and negative elements separately'
  	    usage()
  	    sys.exit(2)	    

!     if(info_filename == None):
  	    print 'ERROR: Need to specify info file'
  	    usage()
--- 112,121 ----

!     if show_separate and boost_filename==None:
  	    print 'ERROR: Need to specify score and margin file if showing positive and negative elements separately'
  	    usage()
  	    sys.exit(2)	    

!     if info_filename == None:
  	    print 'ERROR: Need to specify info file'
  	    usage()
***************
*** 100,115 ****
      outlines = []
      for line in data:
! 	    out = ''
! 	    out += line['iter'] + ' '
! 	    out += line['train'] + ' '
! 	    out += line['test'] + ' '
  	    if (showbound):
  		    out += line['bound'] + ' '
  	    outlines.append(out+'\n')
- 	    
      f = open('error.dat', 'w')
      f.writelines(outlines)
      f.close()

      outlines = []

--- 152,173 ----
      outlines = []
      for line in data:
! 	    out = line['iter'] + ' ' + line['train'] + ' ' + line['test'] + ' '
  	    if (showbound):
  		    out += line['bound'] + ' '
  	    outlines.append(out+'\n')
      f = open('error.dat', 'w')
      f.writelines(outlines)
      f.close()

+     if show_separate:
+ 	    lines = getErrors(boost_filename)
+ 	    outlines = []
+ 	    for line in lines:
+ 		    out = str(line['iter']) + ' ' + str(line['neg_err']) + ' ' + str(line['pos_err']) + ' ' + str(line['err'])
+ 		    outlines.append(out+'\n')
+ 	    f = open('error_sep.dat', 'w')
+ 	    f.writelines(outlines)
+ 	    f.close()
+ 
      outlines = []

***************
*** 119,129 ****
      outlines.append('set ylabel "Error"\n')
      outlines.append('set title "' + info_filename + ' Error"\n')
      if (logaxis):
  	    outlines.append('set logscale x 10\n')
      out = ''
!     out += 'plot "error.dat" using 1:2 title \'train\' with lines'
!     out += ', "error.dat" using 1:3 title \'test\' with lines'
!     if showbound:
! 	    out += ', "error.dat" using 1:4 title \'bound\' with lines'
      out += '\n'
      outlines.append(out)
--- 177,197 ----
      outlines.append('set ylabel "Error"\n')
      outlines.append('set title "' + info_filename + ' Error"\n')
+     if show_separate:
+ 	    outlines.append('set title "' + info_filename + ' Error (POSITIVE EXAMPLE ERROR MAY BE NEGATIVE EXAMPLE ERROR)"\n')
      if (logaxis):
  	    outlines.append('set logscale x 10\n')
+     outlines.append('set style line 10 lt 1 lw 1 pt 5 ps 0.65\n')
+     outlines.append('set style line 11 lt 3 lw 1 pt 1 ps 0.65\n')
+     outlines.append('set style line 12 lt 5 lw 1 pt 1 ps 0.65\n')
      out = ''
!     if show_separate:
! 	    out += 'plot "error_sep.dat" using 1:2 title \'Negative Example Error\' with lines linestyle 10'
! 	    out += ', "error_sep.dat" using 1:3 title \'Positive Example Error\' with lines linestyle 11'
! 	    out += ', "error_sep.dat" using 1:4 title \'Error\' with lines linestyle 12'
!     else:
! 	    out += 'plot "error.dat" using 1:2 title \'train\' with lines linestyle 10'
! 	    out += ', "error.dat" using 1:3 title \'test\' with lines linestyle 11'
! 	    if showbound:
! 		    out += ', "error.dat" using 1:4 title \'bound\' with lines'
      out += '\n'
      outlines.append(out)
***************
*** 136,148 ****
      os.system('gnuplot ' + pngfilename)

- #####################################
-     #    Now do scores file for separate option
- #####################################
-     
-     if show_separate:
- 	    do_separate()
- 
- 
- 

  if __name__ == "__main__":
--- 204,207 ----