d@0: #!/usr/bin/python
d@0: 
d@0: import sys
d@0: import os
d@0: import csv
d@0: import matplotlib.pyplot as plt
d@0: import numpy as np
d@0: import scipy as sp
d@0: import scipy.stats
d@0: 
d@0: # CONFIGURATION
d@0: 
d@0: # Which type(s) of plot do you want? 
d@0: enable_boxplot    = True      # show box plot
d@0: enable_confidence = False     # show confidence interval
d@0: confidence        = 0.90      # confidence value (for confidence interval plot)
d@0: enable_individual = False     # show all individual ratings
d@0: show_individual   = []        # show specific individuals
d@0: show_legend       = False     # show names of individuals
d@0: #TODO: Merge, implement this functionality
d@0: #TODO: Control by CLI arguments (plot types, save and/or show, ...) 
d@0: 
d@0: # Enter folder where rating CSV files are (generated with score_parser.py or same format).
d@0: rating_folder = '../saves/ratings/' # folder with rating csv files
d@0: 
d@0: # Font settings
d@0: font = {'weight' : 'bold',
d@0:         'size'   : 10}
d@0: plt.rc('font', **font)
d@0: 
d@0: 
d@0: # CODE
d@0: 
d@0: # get every csv file in folder
d@0: for file in os.listdir(rating_folder): # You have to put this in folder where rating csv files are.
d@0:     if file.endswith(".csv"):
d@0:         page_name = file[:-4] # file name (without extension) is page ID
d@0: 
d@0:         # get header
d@0:         with open(rating_folder+file, 'rb') as readfile: # read this csv file
d@0:             filereader = csv.reader(readfile, delimiter=',')
d@0:             headerrow = filereader.next() # use headerrow as X-axis
d@0:             headerrow = headerrow[1:]
d@0: 
d@0:         # read ratings into matrix
d@0: #         ratings = np.loadtxt(open(rating_folder+file,"rb"),
d@0: #                             delimiter=",",
d@0: #                             skiprows=1,
d@0: #                             usecols=range(1,len(headerrow)+1)
d@0: #                             )
d@0:             ratings = np.genfromtxt(readfile,
d@0:                                    delimiter=",",
d@0:                                    #skip_header = 1,
d@0:                                    converters = {3: lambda s: float(s or 'Nan')},
d@0:                                    usecols=range(1,len(headerrow)+1)
d@0:                                    )
d@0:         
d@0:         # assert at least 2 subjects (move on to next file if violated)
d@0:         if ratings.shape[0]<2:
d@0:             print "WARNING: Just one subject for " + page_name + ". Moving on to next file."
d@0:             break
d@0: 
d@0:         # BOXPLOT
d@0:         if enable_boxplot:
d@0:             plt.boxplot(ratings)
d@0:             
d@0:         # CONFIDENCE INTERVAL
d@0:         if enable_confidence:
d@0:             iterator = 0
d@0:             for column in ratings.T: # iterate over transposed matrix
d@0:                 # remove all 'Nan's from column
d@0:                 column = column[~np.isnan(column)]
d@0:             
d@0:                 # get number of non-Nan ratings (= #subjects)
d@0:                 n = column.size
d@0:         
d@0:                 # get mean
d@0:                 mean_rating = np.mean(column)
d@0:         
d@0:                 # get errors
d@0:                 err = scipy.stats.sem(column)* sp.stats.t._ppf((1+confidence)/2., n-1)
d@0:         
d@0:                 # draw plot
d@0:                 plt.errorbar(iterator+1, 
d@0:                             mean_rating, 
d@0:                             yerr=err,
d@0:                             marker="x",
d@0:                             color ="k",
d@0:                             markersize=12,
d@0:                             linestyle='None')
d@0:                             
d@0:                 iterator += 1 # increase counter
d@0:     
d@0:     
d@0:         # INDIVIDUAL PLOT
d@0:         if enable_individual or show_individual:
d@0:             # marker list and color map to cycle through
d@0:             markerlist = ["x", ".", "o", "*", "+", "v", ">", "<", "8", "s", "p"]
d@0:             colormap = ['b', 'r', 'g', 'c', 'm', 'y', 'k'] 
d@0:             increment = 0
d@0:             linehandles = []
d@0:             legendnames = []
d@0:             with open(rating_folder+file, 'rb') as readfile: # read this csv file
d@0:                 filereader = csv.reader(readfile, delimiter=',')
d@0:                 headerrow = filereader.next() # use headerrow as X-axis
d@0:                 headerrow = headerrow[1:]
d@0:                 for row in filereader:
d@0:                     subject_id = row[0][:-4] # read from beginning of line
d@0:                     # assume plotting all individuals if no individual(s) specified
d@0:                     if not show_individual or subject_id in show_individual:
d@0:                         plothandle, = plt.plot(range(1,len(row)), # x-values
d@0:                                  ratings[increment,:],#row[1:], # y-values: csv values except subject name
d@0:                                  color=colormap[increment%len(colormap)],
d@0:                                  marker=markerlist[increment%len(markerlist)],
d@0:                                  markersize=10,
d@0:                                  linestyle='None',
d@0:                                  label=subject_id
d@0:                                 )
d@0:                         linehandles.append(plothandle)
d@0:                         legendnames.append(subject_id)
d@0:                         if show_legend:
d@0:                             plt.legend(linehandles, legendnames,
d@0:                                        loc='upper right',
d@0:                                        bbox_to_anchor=(1.1, 1),
d@0:                                        borderaxespad=0.,
d@0:                                        numpoints=1 # remove extra marker
d@0:                                        )
d@0:                     increment += 1 # increase counter
d@0: 
d@0:         # TITLE, AXIS LABELS AND LIMITS
d@0:         plt.title(page_name)
d@0:         plt.xlabel('Fragment')
d@0:         plt.xlim(0, len(headerrow)+1) # only show relevant region, leave space left & right)
d@0:         plt.xticks(range(1, len(headerrow)+1), headerrow) # show fragment names
d@0:         plt.ylabel('Rating')
d@0:         plt.ylim(0,1)
d@0:         
d@0:         
d@0: 
d@0:         # SHOW PLOT
d@0:         #plt.show()
d@0:         #exit()
d@0:     
d@0:         # SAVE PLOT
d@0:         # automatically 
d@0:         plot_type = ("-box" if enable_boxplot else "") + \
d@0:                     ("-conf" if enable_confidence else "") + \
d@0:                     ("-ind" if enable_individual else "")
d@0:         plt.savefig(rating_folder+page_name+plot_type+".png")
d@0:         plt.close()