tomwalters@423: #!/usr/bin/python
tomwalters@423: """
tomwalters@423: plot_munged_results.py
tomwalters@423: 
tomwalters@423: """
tomwalters@423: 
tom@440: import matplotlib as mpl
tom@440: mpl.use('PDF')
tomwalters@423: import numpy as np
tomwalters@423: import pylab as p
tom@440: #import matplotlib.pyplot as plt
tomwalters@425: 
tomwalters@423: f=open("results_test_all.csv","r")
tomwalters@423: results = dict()
tomwalters@423: for line in f:
tomwalters@423:   if line[0] != "#":
tomwalters@423:     values = line.strip().split(",")
tomwalters@423:     results.setdefault(values[3],dict())
tomwalters@423:     results[values[3]].setdefault(values[0], dict())
tomwalters@423:     results[values[3]][values[0]].setdefault(values[1], dict())
tomwalters@424:     results[values[3]][values[0]][values[1]].setdefault(int(values[4]), dict())
tomwalters@424:     results[values[3]][values[0]][values[1]][int(values[4])].setdefault(int(values[5]), dict())
tomwalters@424:     results[values[3]][values[0]][values[1]][int(values[4])][int(values[5])].setdefault(int(values[6]), dict())
tomwalters@423:     if values[2] == 'clean':
tomwalters@424:        snr = 50
tom@440:        results[values[3]][values[0]][values[1]][int(values[4])][int(values[5])][int(values[6])][snr] = float(values[7])
tomwalters@423:     else:
tomwalters@423:        snr = int(values[2])
tomwalters@425:        results[values[3]][values[0]][values[1]][int(values[4])][int(values[5])][int(values[6])][snr] = float(values[7])
tomwalters@423: #    results[values[3]].append((values[1],values[2],values[2],values[4]))
tomwalters@423: 
tom@440: ax = mpl.pyplot.subplot(111)
tomwalters@423: train_set = 'inner'
tom@440: for hmm_iterations in [2,3,15]:
tom@440:   for hmm_states in [3,4]:
tom@440:     for hmm_components in [3,4]:
tom@440:       lines = []
tom@440:       labels = []
tom@440:       ax.cla()
tom@440:       for feature_type in ('mfcc', 'mfcc_vtln', 'aim'):
tom@440:         for feature_subtype in results[train_set][feature_type].keys():
tom@440:           try:
tom@440:             this_line = results[train_set][feature_type][feature_subtype][hmm_states][hmm_components][hmm_iterations].items()
tom@440:             this_line.sort(cmp=lambda x,y: x[0] - y[0])
tom@440:             xs, ys = zip(*this_line)
tom@440:             xs = list(xs)
tom@440:             ys = list(ys)
tom@440:             line, = ax.plot(xs,ys,'-o',linewidth=2)
tom@440:             lines.append(line)
tom@440:             labels.append(feature_type + "_" + feature_subtype)
tom@440:           except KeyError:
tom@440:             print "Data not found"
tom@440:       p.legend(lines, labels, 'upper left', shadow=True)
tom@440:       p.xlabel('SNR/dB')
tom@440:       p.ylabel('Recognition performance %')
tom@440:       output_file = ("recognition_vs_snr_%diterations_%dstates_%d_components.pdf" % (hmm_iterations, hmm_states, hmm_components))
tom@440:       p.savefig(output_file)