Mercurial > hg > webaudioevaluationtool
changeset 241:a4b5246ffb32
Scripts: added 'confidence interval plot' (rigorous checking needed) and minor fixes to other scripts.
| author | Brecht De Man <b.deman@qmul.ac.uk> |
|---|---|
| date | Thu, 25 Jun 2015 17:14:33 +0100 |
| parents | 16737f53817c |
| children | e08f2b155d8b |
| files | .hgignore.orig scripts/comment_parser.py scripts/score_boxplot.py scripts/score_confidence.py |
| diffstat | 4 files changed, 157 insertions(+), 31 deletions(-) [+] |
line wrap: on
line diff
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/.hgignore.orig Thu Jun 25 17:14:33 2015 +0100 @@ -0,0 +1,28 @@ +syntax: glob +.project/** +.project +docs/SMC15/smc2015.log +docs/SMC15/smc2015template.aux +docs/SMC15/smc2015template.blg +docs/SMC15/smc2015template.log +docs/SMC15/smc2015template.out +docs/SMC15/smc2015template.pdf +docs/SMC15/smc2015template.synctex.gz +re:^docs/ExperimentVariableControl\.aux$ +re:^docs/ExperimentVariableControl\.log$ +re:^docs/ExperimentVariableControl\.synctex\.gz$ +*.aux +*.bbl +*.gz +*.log +*.blg +*.DS_Store +re:^RealismEval/ +re:^SynthReal/ +re:^SynthRealismTest\.html$ +re:^example_eval/paperExample\.xml$ +re:^example_eval/project\.xml\.orig$ +re:^index2\.html$ +re:^realismTest\.html$ +saves/*.xml +saves/timelines \ No newline at end of file
--- a/scripts/comment_parser.py Thu Jun 25 16:40:39 2015 +0100 +++ b/scripts/comment_parser.py Thu Jun 25 17:14:33 2015 +0100 @@ -1,4 +1,5 @@ #!/usr/bin/python +# -*- coding: utf-8 -*- import xml.etree.ElementTree as ET import os @@ -22,7 +23,7 @@ os.makedirs(page_name) # for page [page_name], print comments related to fragment [id] - for audioelement in root.findall("*/[@id='"+page_name+"']/audioelement"): + for audioelement in root.findall("*/[@id='"+page_name+"']/audioelement"): #TODO in audioholder.findall(...) if audioelement is not None: # Check it exists audio_id = str(audioelement.get('id')) @@ -44,9 +45,9 @@ writer.writerow(['']) else: # anonymous comments: - writer.writerow([commentstr]) + writer.writerow([commentstr.encode("utf-8")]) # comments with (file) name: - #writer.writerow([file[:-4]] + [commentstr]) + #writer.writerow([file[:-4]] + [commentstr.encode("utf-8")]) #TODO Replace 'new line' in comment with something else?
--- a/scripts/score_boxplot.py Thu Jun 25 16:40:39 2015 +0100 +++ b/scripts/score_boxplot.py Thu Jun 25 17:14:33 2015 +0100 @@ -4,8 +4,11 @@ import matplotlib.pyplot as plt import numpy as np +# Enter folder where rating CSV files are (generated with score_parser.py or same format). +# Add subject names of individual ratings to be marked in 'show_individual'. + rating_folder = 'ratings/' # folder with rating csv files -show_individual = 'frank' +show_individual = [] # add name/list of names of individuals to plot # get every csv file in folder for file in os.listdir(rating_folder): # You have to put this in folder where rating csv files are. @@ -27,33 +30,34 @@ # draw boxplot plt.boxplot(ratings) - - # add rating of individual(s) - with open(rating_folder+file, 'r') as readfile: # read this csv file - filereader = csv.reader(readfile, delimiter=',') - headerrow = filereader.next() # use headerrow as X-axis - headerrow = headerrow[1:] - markerlist = ["x", ".", "o", "*", "+", "v", ">", "<", "8", "s", "p"] - increment = 0 - linehandles = [] - legendnames = [] - for row in filereader: - subject_id = row[0][:-4] - if subject_id in show_individual: - plothandle, = plt.plot(range(1,len(row)), # x-values - row[1:], # y-values: csv values except subject name - color='k', - marker=markerlist[increment%len(markerlist)], - markersize=10, - linestyle='None', - label=subject_id - ) - increment += 1 # increase counter - linehandles.append(plothandle) - legendnames.append(subject_id) - plt.legend(linehandles, legendnames, - loc='upper right', - bbox_to_anchor=(1.1, 1), borderaxespad=0.) + + if not show_individual: + # add rating of individual(s) + with open(rating_folder+file, 'r') as readfile: # read this csv file + filereader = csv.reader(readfile, delimiter=',') + headerrow = filereader.next() # use headerrow as X-axis + headerrow = headerrow[1:] + markerlist = ["x", ".", "o", "*", "+", "v", ">", "<", "8", "s", "p"] + increment = 0 + linehandles = [] + legendnames = [] + for row in filereader: + subject_id = row[0][:-4] + if subject_id in show_individual: + plothandle, = plt.plot(range(1,len(row)), # x-values + row[1:], # y-values: csv values except subject name + color='k', + marker=markerlist[increment%len(markerlist)], + markersize=10, + linestyle='None', + label=subject_id + ) + increment += 1 # increase counter + linehandles.append(plothandle) + legendnames.append(subject_id) + plt.legend(linehandles, legendnames, + loc='upper right', + bbox_to_anchor=(1.1, 1), borderaxespad=0.) plt.xlabel('Fragment')
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/scripts/score_confidence.py Thu Jun 25 17:14:33 2015 +0100 @@ -0,0 +1,93 @@ +import sys +import os +import csv +import matplotlib.pyplot as plt +import numpy as np +import scipy as sp +import scipy.stats + +# Enter folder where rating CSV files are (generated with score_parser.py or same format). +# Add subject names of individual ratings to be marked in 'show_individual'. +# Choose confidence value. + +rating_folder = 'ratings/' # folder with rating csv files +show_individual = [] # add name/list of names of individuals to plot +confidence = .9 # confidence percentage (usually 80%-99%) + +# get every csv file in folder +for file in os.listdir(rating_folder): # You have to put this in folder where rating csv files are. + if file.endswith(".csv"): + page_name = file[:-4] # file name (without extension) is page ID + + # get header + with open(rating_folder+file, 'r') as readfile: # read this csv file + filereader = csv.reader(readfile, delimiter=',') + headerrow = filereader.next() # use headerrow as X-axis + headerrow = headerrow[1:] + + # read ratings into matrix + ratings = np.loadtxt(open(rating_folder+file,"rb"), + delimiter=",", + skiprows=1, + usecols=range(1,len(headerrow)+1) + ) + + # get number of rows (= subjects) + n = ratings.shape[1] + + # get means + means = np.mean(ratings, axis=0) + + # get errors + err = scipy.stats.sem(ratings)* sp.stats.t._ppf((1+confidence)/2., n-1) + + # draw plot + plt.errorbar(range(1,len(headerrow)+1), + means, + yerr=err, + marker="x", + markersize=10, + linestyle='None') + + if not show_individual: + # add rating of individual(s) + with open(rating_folder+file, 'r') as readfile: # read this csv file + filereader = csv.reader(readfile, delimiter=',') + headerrow = filereader.next() # use headerrow as X-axis + headerrow = headerrow[1:] + markerlist = ["x", ".", "o", "*", "+", "v", ">", "<", "8", "s", "p"] + increment = 0 + linehandles = [] + legendnames = [] + for row in filereader: + subject_id = row[0][:-4] + if subject_id in show_individual: + plothandle, = plt.plot(range(1,len(row)), # x-values + row[1:], # y-values: csv values except subject name + color='k', + marker=markerlist[increment%len(markerlist)], + markersize=10, + linestyle='None', + label=subject_id + ) + increment += 1 # increase counter + linehandles.append(plothandle) + legendnames.append(subject_id) + plt.legend(linehandles, legendnames, + loc='upper right', + bbox_to_anchor=(1.1, 1), borderaxespad=0.) + + + plt.xlabel('Fragment') + plt.title('Confidence interval '+page_name) + plt.xlim(0, len(headerrow)+1) # only show relevant region, leave space left & right) + plt.xticks(range(1, len(headerrow)+1), headerrow) # show fragment names + + plt.ylabel('Rating') + plt.ylim(0,1) + + #plt.show() # show plot + #exit() + + plt.savefig(rating_folder+page_name+"-conf.png") + plt.close()