API tools faq
paste
Advertisement
wagner-cipriano

plot a pretty confusion matrix with seaborn py like matlab

wagner-cipriano
Jul 3rd, 2018
279
0
Never
Add comment
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
Python 9.47 KB | None | 0 0
raw download clone embed print report
  1. # -*- coding: utf-8 -*-
  2. """
  3. plot a pretty confusion matrix with seaborn
  4. works in python 2 and 3
  5.  
  6. Created on Mon Jun 25 14:17:37 2018
  7. @author: Wagner Cipriano - wagnerbhbr-gmail
  8. REFerences:
  9.  https://www.mathworks.com/help/nnet/ref/plotconfusion.html
  10.    https://www.mathworks.com/help/examples/nnet/win64/PlotConfusionMatrixUsingCategoricalLabelsExample_02.png
  11.  https://stackoverflow.com/questions/28200786/how-to-plot-scikit-learn-classification-report
  12.  https://stackoverflow.com/questions/5821125/how-to-plot-confusion-matrix-with-string-axis-rather-than-integer-in-python
  13.  https://www.programcreek.com/python/example/96197/seaborn.heatmap
  14.  https://stackoverflow.com/questions/19233771/sklearn-plot-confusion-matrix-with-labels/31720054
  15. """
  16.  
  17. #imports
  18. from pandas import DataFrame
  19. import numpy as np
  20. import matplotlib.pyplot as plt
  21. import matplotlib.font_manager as fm
  22. from matplotlib.collections import QuadMesh
  23. import seaborn as sn
  24.  
  25.  
  26. def get_new_fig(fn, figsize=[9,9]):
  27.     ## Init graphics
  28.     fig1 = plt.figure(fn, figsize)
  29.     ax1 = fig1.gca()   #Get Current Axis
  30.     ax1.cla() # clear existing plot
  31.     return fig1, ax1
  32. #
  33.  
  34. def configcell_text_and_colors(array_df, lin, col, oText, facecolors, posi, fz, fmt, show_null_values=0):
  35.     """
  36.      config cell text and colors
  37.      and return text elements to add and to dell
  38.      @TODO: use fmt
  39.    """
  40.     text_add = []; text_del = [];
  41.     cell_val = array_df[lin][col]
  42.     tot_all = array_df[-1][-1]
  43.     per = (float(cell_val) / tot_all) * 100
  44.     curr_column = array_df[:,col]
  45.     ccl = len(curr_column)
  46.  
  47.     #last line  and/or last column
  48.     if(col == (ccl - 1)) or (lin == (ccl - 1)):
  49.         #tots and percents
  50.         if(cell_val != 0):
  51.             if(col == ccl - 1) and (lin == ccl - 1):
  52.                 tot_rig = 0
  53.                 for i in range(array_df.shape[0] - 1):
  54.                     tot_rig += array_df[i][i]
  55.                 per_ok = (float(tot_rig) / cell_val) * 100
  56.             elif(col == ccl - 1):
  57.                 tot_rig = array_df[lin][lin]
  58.                 per_ok = (float(tot_rig) / cell_val) * 100
  59.             elif(lin == ccl - 1):
  60.                 tot_rig = array_df[col][col]
  61.                 per_ok = (float(tot_rig) / cell_val) * 100
  62.             per_err = 100 - per_ok
  63.         else:
  64.             per_ok = per_err = 0
  65.  
  66.         per_ok_s = ['%.2f%%'%(per_ok), '100%'] [per_ok == 100]
  67.  
  68.         #text to DEL
  69.         text_del.append(oText)
  70.  
  71.         #text to ADD
  72.         font_prop = fm.FontProperties(weight='bold', size=fz)
  73.         text_kwargs = dict(color='w', ha="center", va="center", gid='sum', fontproperties=font_prop)
  74.         lis_txt = ['%d'%(cell_val), per_ok_s, '%.2f%%'%(per_err)]
  75.         lis_kwa = [text_kwargs]
  76.         dic = text_kwargs.copy(); dic['color'] = 'g'; lis_kwa.append(dic);
  77.         dic = text_kwargs.copy(); dic['color'] = 'r'; lis_kwa.append(dic);
  78.         lis_pos = [(oText._x, oText._y-0.3), (oText._x, oText._y), (oText._x, oText._y+0.3)]
  79.         for i in range(len(lis_txt)):
  80.             newText = dict(x=lis_pos[i][0], y=lis_pos[i][1], text=lis_txt[i], kw=lis_kwa[i])
  81.             #print 'lin: %s, col: %s, newText: %s' %(lin, col, newText)
  82.             text_add.append(newText)
  83.         #print '\n'
  84.  
  85.         #set background color for sum cells (last line and last column)
  86.         carr = [0.27, 0.30, 0.27, 1.0]
  87.         if(col == ccl - 1) and (lin == ccl - 1):
  88.             carr = [0.17, 0.20, 0.17, 1.0]
  89.         facecolors[posi] = carr
  90.  
  91.     else:
  92.         if(per > 0):
  93.             txt = '%s\n%.2f%%' %(cell_val, per)
  94.         else:
  95.             if(show_null_values == 0):
  96.                 txt = ''
  97.             elif(show_null_values == 1):
  98.                 txt = '0'
  99.             else:
  100.                 txt = '0\n0.0%'
  101.         oText.set_text(txt)
  102.  
  103.         #main diagonal
  104.         if(col == lin):
  105.             #set color of the textin the diagonal to white
  106.             oText.set_color('w')
  107.             # set background color in the diagonal to blue
  108.             facecolors[posi] = [0.35, 0.8, 0.55, 1.0]
  109.         else:
  110.             oText.set_color('r')
  111.  
  112.     return text_add, text_del
  113. #
  114.  
  115. def insert_totals(df_cm):
  116.     """ insert total column and line (the last ones) """
  117.     sum_col = []
  118.     for c in df_cm.columns:
  119.         sum_col.append( df_cm[c].sum() )
  120.     sum_lin = []
  121.     for item_line in df_cm.iterrows():
  122.         sum_lin.append( item_line[1].sum() )
  123.     df_cm['sum_lin'] = sum_lin
  124.     sum_col.append(np.sum(sum_lin))
  125.     df_cm.loc['sum_col'] = sum_col
  126.     #print ('\ndf_cm:\n', df_cm, '\n\b\n')
  127. #
  128.  
  129.  
  130.  
  131. def pretty_plot_confusion_matrix(df_cm, annot=True, cmap="Oranges", fmt='.2f', fz=11,
  132.       lw=0.5, cbar=False, figsize=[8,8], show_null_values=0):
  133.     """
  134.      print conf matrix with default layout (like matlab)
  135.      params:
  136.        df_cm   dataframe (pandas) without totals
  137.        annot   print text in each cell
  138.        cmap    Oranges,Oranges_r,YlGnBu,Blues,RdBu, ... see:
  139.        fz      fontsize
  140.        lw      linewidth
  141.    """
  142.     # create "Total" column
  143.     insert_totals(df_cm)
  144.  
  145.     #this is for print allways in the same window
  146.     fig, ax1 = get_new_fig('Conf matrix default', figsize)
  147.  
  148.     #thanks for seaborn
  149.     ax = sn.heatmap(df_cm, annot=annot, annot_kws={"size": fz}, linewidths=lw, ax=ax1,
  150.                     cbar=cbar, cmap=cmap, linecolor='w', fmt=fmt)
  151.  
  152.     #set ticklabels rotation
  153.     ax.set_xticklabels(ax.get_xticklabels(), rotation = 45, fontsize = 10)
  154.     ax.set_yticklabels(ax.get_yticklabels(), rotation = 25, fontsize = 10)
  155.  
  156.     # Turn off all the ticks
  157.     for t in ax.xaxis.get_major_ticks():
  158.         t.tick1On = False
  159.         t.tick2On = False
  160.     for t in ax.yaxis.get_major_ticks():
  161.         t.tick1On = False
  162.         t.tick2On = False
  163.  
  164.     #face colors list
  165.     quadmesh = ax.findobj(QuadMesh)[0]
  166.     facecolors = quadmesh.get_facecolors()
  167.  
  168.     #iter in text elements
  169.     array_df = np.array( df_cm.to_records(index=False).tolist() )
  170.     text_add = []; text_del = [];
  171.     posi = -1 #from left to right, bottom to top.
  172.     for t in ax.collections[0].axes.texts: #ax.texts:
  173.         pos = np.array( t.get_position()) - [0.5,0.5]
  174.         lin = int(pos[1]); col = int(pos[0]);
  175.         posi += 1
  176.         #print ('>>> pos: %s, posi: %s, val: %s, txt: %s' %(pos, posi, array_df[lin][col], t.get_text()))
  177.  
  178.         #set text
  179.         txt_res = configcell_text_and_colors(array_df, lin, col, t, facecolors, posi, fz, fmt, show_null_values)
  180.  
  181.         text_add.extend(txt_res[0])
  182.         text_del.extend(txt_res[1])
  183.  
  184.     #remove the old ones
  185.     for item in text_del:
  186.         item.remove()
  187.     #append the new ones
  188.     for item in text_add:
  189.         ax.text(item['x'], item['y'], item['text'], **item['kw'])
  190.  
  191.     #titles and legends
  192.     ax.set_title('Confusion matrix')
  193.     ax.set_xlabel('Predicted')
  194.     ax.set_ylabel('Actual')
  195.     plt.tight_layout()  #set layout slim
  196.     plt.show()
  197. #
  198.  
  199. def plot_confusion_matrix_from_data(y_test, predictions, columns=None, annot=True,
  200.       cmap="Oranges", fmt='.2f', fz=11, lw=0.5, cbar=False, figsize=[8,8], show_null_values=0):
  201.     """
  202.        plot confusion matrix function with y_test (actual values) and predictions (predic),
  203.        whitout a confusion matrix yet
  204.    """
  205.     from sklearn.metrics import confusion_matrix
  206.     from pandas import DataFrame
  207.  
  208.     #data
  209.     if(not columns):
  210.         columns = range(1, len(np.unique(y_test))+1)
  211.  
  212.     confm = confusion_matrix(y_test, predictions)
  213.     cmap = 'Oranges';
  214.     fz = 11;
  215.     figsize=[9,9];
  216.     show_null_values = 2
  217.     df_cm = DataFrame(confm, index=columns, columns=columns)
  218.     pretty_plot_confusion_matrix(df_cm, fz=fz, cmap=cmap, figsize=figsize, show_null_values=show_null_values)
  219. #
  220.  
  221.  
  222.  
  223. #
  224. #Test functions
  225. #
  226. def _test_cm():
  227.     #test function with confusion matrix done
  228.     array = np.array( [[13,  0,  1,  0,  2,  0],
  229.                        [ 0, 50,  2,  0, 10,  0],
  230.                        [ 0, 13, 16,  0,  0,  3],
  231.                        [ 0,  0,  0, 13,  1,  0],
  232.                        [ 0, 40,  0,  1, 15,  0],
  233.                        [ 0,  0,  0,  0,  0, 20]])
  234.     #get pandas dataframe
  235.     df_cm = DataFrame(array, index=range(1,7), columns=range(1,7))
  236.     #colormap: see this and choose your more dear
  237.     cmap = 'PuRd'
  238.     pretty_plot_confusion_matrix(df_cm, cmap=cmap)
  239. #
  240.  
  241. def _test_data_class():
  242.     """ test function with y_test (actual values) and predictions (predic) """
  243.     #data
  244.     y_test = np.array([1,2,3,4,5, 1,2,3,4,5, 1,2,3,4,5, 1,2,3,4,5, 1,2,3,4,5, 1,2,3,4,5, 1,2,3,4,5, 1,2,3,4,5, 1,2,3,4,5, 1,2,3,4,5, 1,2,3,4,5, 1,2,3,4,5, 1,2,3,4,5, 1,2,3,4,5, 1,2,3,4,5, 1,2,3,4,5, 1,2,3,4,5, 1,2,3,4,5, 1,2,3,4,5])
  245.     predic = np.array([1,2,4,3,5, 1,2,4,3,5, 1,2,3,4,4, 1,4,3,4,5, 1,2,4,4,5, 1,2,4,4,5, 1,2,4,4,5, 1,2,4,4,5, 1,2,3,3,5, 1,2,3,3,5, 1,2,3,4,4, 1,2,3,4,1, 1,2,3,4,1, 1,2,3,4,1, 1,2,4,4,5, 1,2,4,4,5, 1,2,4,4,5, 1,2,4,4,5, 1,2,3,4,5])
  246.     columns = []
  247.     annot = True;
  248.     cmap = 'Oranges';
  249.     fmt = '.2f'
  250.     lw = 0.5
  251.     cbar = False
  252.     show_null_values = 2
  253.     #size::
  254.     fz = 12;
  255.     figsize = [9,9];
  256.     if(len(y_test) > 10):
  257.         fz=9; figsize=[14,14];
  258.     plot_confusion_matrix_from_data(y_test, predic, columns,
  259.       annot, cmap, fmt, fz, lw, cbar, figsize, show_null_values)
  260. #
  261.  
  262.  
  263. #
  264. #MAIN
  265. #
  266. if(__name__ == '__main__'):
  267.     print('__main__')
  268.     print('_test_cm: test function with confusion matrix done\nand pause')
  269.     _test_cm()
  270.     plt.pause(5)
  271.     print('_test_data_class: test function with y_test (actual values) and predictions (predic)')
  272.     _test_data_class()
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement
Public Pastes
Advertisement
We use cookies for various purposes including analytics. By continuing to use Pastebin, you agree to our use of cookies as described in the Cookies Policy.  OK, I Understand
Not a member of Pastebin yet?
Sign Up, it unlocks many cool features!