-
Notifications
You must be signed in to change notification settings - Fork 5
/
Copy pathconfusion_matrix_pretty_print.py
308 lines (274 loc) · 11.1 KB
/
confusion_matrix_pretty_print.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
# -*- coding: utf-8 -*-
"""
plot a pretty confusion matrix with seaborn
Created on Mon Jun 25 14:17:37 2018
@author: Wagner Cipriano - wagnerbhbr - gmail - CEFETMG / MMC
REFerences:
https://www.mathworks.com/help/nnet/ref/plotconfusion.html
https://stackoverflow.com/questions/28200786/how-to-plot-scikit-learn-classification-report
https://stackoverflow.com/questions/5821125/how-to-plot-confusion-matrix-with-string-axis-rather-than-integer-in-python
https://www.programcreek.com/python/example/96197/seaborn.heatmap
https://stackoverflow.com/questions/19233771/sklearn-plot-confusion-matrix-with-labels/31720054
http://scikit-learn.org/stable/auto_examples/model_selection/plot_confusion_matrix.html#sphx-glr-auto-examples-model-selection-plot-confusion-matrix-py
"""
#imports
from pandas import DataFrame
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.font_manager as fm
from matplotlib.collections import QuadMesh
import seaborn as sn
def get_new_fig(fn, figsize=[9,9]):
""" Init graphics """
fig1 = plt.figure(fn, figsize)
ax1 = fig1.gca() #Get Current Axis
ax1.cla() # clear existing plot
return fig1, ax1
#
def configcell_text_and_colors(array_df, lin, col, oText, facecolors, posi, fz, fmt, show_null_values=0):
"""
config cell text and colors
and return text elements to add and to dell
@TODO: use fmt
"""
text_add = []; text_del = [];
cell_val = array_df[lin][col]
tot_all = array_df[-1][-1]
per = (float(cell_val) / tot_all) * 100
curr_column = array_df[:,col]
ccl = len(curr_column)
#last line and/or last column
if(col == (ccl - 1)) or (lin == (ccl - 1)):
#tots and percents
if(cell_val != 0):
if(col == ccl - 1) and (lin == ccl - 1):
tot_rig = 0
for i in range(array_df.shape[0] - 1):
tot_rig += array_df[i][i]
per_ok = (float(tot_rig) / cell_val) * 100
elif(col == ccl - 1):
tot_rig = array_df[lin][lin]
per_ok = (float(tot_rig) / cell_val) * 100
elif(lin == ccl - 1):
tot_rig = array_df[col][col]
per_ok = (float(tot_rig) / cell_val) * 100
per_err = 100 - per_ok
else:
per_ok = per_err = 0
per_ok_s = ['%.2f%%'%(per_ok), '100%'] [per_ok == 100]
#text to DEL
text_del.append(oText)
#text to ADD
font_prop = fm.FontProperties(weight='bold', size=fz)
text_kwargs = dict(color='w', ha="center", va="center", gid='sum', fontproperties=font_prop)
lis_txt = ['%d'%(cell_val), per_ok_s, '%.2f%%'%(per_err)]
lis_kwa = [text_kwargs]
dic = text_kwargs.copy(); dic['color'] = 'g'; lis_kwa.append(dic);
dic = text_kwargs.copy(); dic['color'] = 'r'; lis_kwa.append(dic);
lis_pos = [(oText._x, oText._y-0.3), (oText._x, oText._y), (oText._x, oText._y+0.3)]
for i in range(len(lis_txt)):
newText = dict(x=lis_pos[i][0], y=lis_pos[i][1], text=lis_txt[i], kw=lis_kwa[i])
#print ('lin: %s, col: %s, newText: %s' %(lin, col, newText))
text_add.append(newText)
#print '\n'
#set background color for sum cells (last line and last column)
carr = [0.0, 0.0, 0.0, 0.34]
if(col == ccl - 1) and (lin == ccl - 1):
carr = [0.0, 0.0, 0.0, 0.42]
facecolors[posi] = carr
else:
if(per > 0):
txt = '%s\n%.2f%%' %(cell_val, per)
else:
if(show_null_values == 0):
txt = ''
elif(show_null_values == 1):
txt = '0'
else:
txt = '0\n0.0%'
oText.set_text(txt)
#main diagonal
if(col == lin):
#set color of the textin the diagonal to white
oText.set_color('w')
# set background color in the diagonal to blue
facecolors[posi] = [0.35, 0.8, 0.55, 1.0]
else:
oText.set_color('r')
return text_add, text_del
#
def insert_totals(df_cm):
""" insert total column and line (the last ones) """
sum_col = []
for c in df_cm.columns:
sum_col.append( df_cm[c].sum() )
sum_lin = []
for item_line in df_cm.iterrows():
sum_lin.append( item_line[1].sum() )
df_cm['sum_row'] = sum_lin
sum_col.append(np.sum(sum_lin))
df_cm.loc['sum_col'] = sum_col
#print ('\ndf_cm:\n', df_cm, '\n\b\n')
#
def pretty_plot_confusion_matrix(df_cm, annot=True, cmap="Oranges", fmt='.2f', fz=11,
lw=0.5, cbar=False, figsize=[8,8], show_null_values=0, pred_val_axis='y', title='Confusion matrix', path='cm.png'):
"""
print conf matrix with default layout (like matlab)
params:
df_cm dataframe (pandas) without totals
annot print text in each cell
cmap Oranges,Oranges_r,YlGnBu,Blues,RdBu, ... see:
fz fontsize
lw linewidth
pred_val_axis where to show the prediction values (x or y axis)
'col' or 'x': show predicted values in columns (x axis) instead lines
'lin' or 'y': show predicted values in lines (y axis)
"""
if(pred_val_axis in ('col', 'x')):
xlbl = 'Predicted'
ylbl = 'Actual'
else:
xlbl = 'Actual'
ylbl = 'Predicted'
df_cm = df_cm.T
# create "Total" column
insert_totals(df_cm)
#this is for print allways in the same window
fig, ax1 = get_new_fig('Conf matrix default', figsize)
#thanks for seaborn
ax = sn.heatmap(df_cm, annot=annot, annot_kws={"size": fz}, linewidths=lw, ax=ax1,
cbar=cbar, cmap=cmap, linecolor='w', fmt=fmt)
#set ticklabels rotation
ax.set_xticklabels(ax.get_xticklabels(), rotation = 60, fontsize = 10, ha='right')
ax.set_yticklabels(ax.get_yticklabels(), rotation = 25, fontsize = 10)
# Turn off all the ticks
for t in ax.xaxis.get_major_ticks():
t.tick1On = False
t.tick2On = False
for t in ax.yaxis.get_major_ticks():
t.tick1On = False
t.tick2On = False
#face colors list
quadmesh = ax.findobj(QuadMesh)[0]
facecolors = quadmesh.get_facecolors()
#iter in text elements
array_df = np.array( df_cm.to_records(index=False).tolist() )
text_add = []; text_del = [];
posi = -1 #from left to right, bottom to top.
for t in ax.collections[0].axes.texts: #ax.texts:
pos = np.array( t.get_position()) - [0.5,0.5]
lin = int(pos[1]); col = int(pos[0]);
posi += 1
#print('>>> pos: %s, posi: %s, val: %s, txt: %s' %(pos, posi, array_df[lin][col], t.get_text()))
#set text
txt_res = configcell_text_and_colors(array_df, lin, col, t, facecolors, posi, fz, fmt, show_null_values)
text_add.extend(txt_res[0])
text_del.extend(txt_res[1])
#remove the old ones
for item in text_del:
item.remove()
#append the new ones
for item in text_add:
ax.text(item['x'], item['y'], item['text'], **item['kw'])
#titles and legends
ax.set_title(title)
ax.set_xlabel(xlbl)
ax.set_ylabel(ylbl)
plt.tight_layout() #set layout slim
plt.savefig(path, facecolor='w')
plt.show(block=False)
plt.close('all')
#
def plot_confusion_matrix_from_data(y_test, predictions, columns=None, annot=True, cmap="Oranges",
fmt='.2f', fz=11, lw=0.5, cbar=False, figsize=[8,8], show_null_values=0, pred_val_axis='lin'):
"""
plot confusion matrix function with y_test (actual values) and predictions (predic),
whitout a confusion matrix yet
"""
from sklearn.metrics import confusion_matrix
from pandas import DataFrame
#data
if(not columns):
#labels axis integer:
##columns = range(1, len(np.unique(y_test))+1)
#labels axis string:
from string import ascii_uppercase
columns = ['class %s' %(i) for i in list(ascii_uppercase)[0:len(np.unique(y_test))]]
confm = confusion_matrix(y_test, predictions)
cmap = 'Oranges';
fz = 11;
figsize=[9,9];
show_null_values = 2
df_cm = DataFrame(confm, index=columns, columns=columns)
pretty_plot_confusion_matrix(df_cm, fz=fz, cmap=cmap, figsize=figsize, show_null_values=show_null_values, pred_val_axis=pred_val_axis)
#
#
#TEST functions
#
def _test_cm():
#test function with confusion matrix done
array = np.array([[ 0, 0, 0, 0, 0, 0, 0, 7, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 25, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 5, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 418, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 2, 0, 0, 0, 0, 0, 0, 2929, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 1, 0, 0, 0, 0, 0, 0, 165, 0, 0, 0],
[ 4, 0, 0, 0, 0, 0, 0, 873, 0, 6, 17]])
#get pandas dataframe
ind = ['B-Beschwerdeführer',
'B-Dritte',
'B-EGMR',
'B-Kommission/Kammer',
'B-Staat',
'I-Beschwerdeführer',
'I-Dritte',
'I-EGMR',
'I-Kommission/Kammer',
'I-Staat',
'O']
df_cm = DataFrame(array, index=ind, columns=ind)
#colormap: see this and choose your more dear
cmap = 'PuRd'
pretty_plot_confusion_matrix(df_cm, cmap=cmap,figsize=[11,11])
#
def _test_data_class():
""" test function with y_test (actual values) and predictions (predic) """
#data
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, 1,2,3,4,5, 1,2,3,4,5, 1,2,3,4,5])
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, 1,2,3,4,5, 1,2,3,4,5, 1,2,3,4,5])
"""
Examples to validate output (confusion matrix plot)
actual: 5 and prediction 1 >> 3
actual: 2 and prediction 4 >> 1
actual: 3 and prediction 4 >> 10
"""
columns = []
annot = True;
cmap = 'Oranges';
fmt = '.2f'
lw = 0.5
cbar = False
show_null_values = 2
pred_val_axis = 'y'
#size::
fz = 12;
figsize = [9,9];
if(len(y_test) > 10):
fz=9; figsize=[14,14];
plot_confusion_matrix_from_data(y_test, predic, columns,
annot, cmap, fmt, fz, lw, cbar, figsize, show_null_values, pred_val_axis)
#
#
#MAIN function
#
if(__name__ == '__main__'):
print('__main__')
print('_test_cm: test function with confusion matrix done\nand pause')
_test_cm()
plt.pause(5)
print('_test_data_class: test function with y_test (actual values) and predictions (predic)')
_test_data_class()