-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathPlot_box_plots.py
216 lines (167 loc) · 10.3 KB
/
Plot_box_plots.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
# =============================================================================
# code to generate tables from results\ .txt files
# =============================================================================
import numpy as np
import os
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
def get_dataframe(path, mix, loss,name,metric):
for i in ([mix]):
for l in ([loss]):
for s in ([10]):
acc=[]
bacc=[]
mse=[]
f1=[]
auc=[]
precision=[]
for f in range(5):
text_file = open(os.path.join(path,f"results_{s}_mix_{i}_loss_{l}_fold_{f}.txt"))
lines = text_file.readlines()
acc.append(float(lines[3].split(":")[1]))
mse.append(float(lines[4].split(":")[1]))
f1.append(float(lines[5].split(":")[1]))
precision.append(float(lines[7].split(":")[1]))
auc.append(float(lines[8].split(":")[1]))
bacc.append(float(lines[9].split(":")[1]))
if metric=="Accuracy (%)":
met=acc
elif metric=="AUC (%)":
met=auc
elif metric=="F1 (%)":
met=f1
else:
pass
dframe_cross=[]
for i in range(5):
dframe_cross.append([name, i, met[i]])
df = pd.DataFrame(dframe_cross,columns=['Model', 'fold_idx', metric])
return df
def get_datafram_MRI(data,name,metric):
dframe_cross=[]
for i in range(5):
dframe_cross.append([name, i, data[i]])
dfm = pd.DataFrame(dframe_cross,columns=['Model', 'fold_idx', metric])
return dfm
# =============================================================================
# Plot the boxplot
# =============================================================================
def get_full_datframe(metric="AUC (%)"):
if metric=="Accuracy (%)":
MRI_t1ce = [0.7632, 0.8421, 0.7568, 0.8108, 0.8378]
MRI_t1ce_flair= [0.7632, 0.7368, 0.7162, 0.6892, 0.7568]
MRI_t1ce_original = [i/100 for i in [57.895, 65.789, 72.973, 67.568, 64.865]]
MRI_t1ce_flair_original = [i/100 for i in [65.789, 65.789, 75.676, 70.270, 64.865]]
elif metric=="AUC (%)":
MRI_t1ce = [0.8565426628, 0.8586716524, 0.8439452704, 0.8714786756,0.8857555328]
MRI_t1ce_flair = [0.8899334357, 0.8579415954, 0.8337867198, 0.7882505612, 0.8338240314]
MRI_t1ce_original = [i/100 for i in [58.640, 66.425, 72.008, 67.794, 66.436]]
MRI_t1ce_flair_original = [i/100 for i in [72.779, 71.202, 66.195, 76.734, 69.757]]
elif metric=="F1 (%)":
MRI_t1ce = [0.7629, 0.8406, 0.7541, 0.8125, 0.8465]
MRI_t1ce_flair= [0.7599, 0.7014, 0.6976, 0.5907, 0.690]
MRI_t1ce_original = [i/100 for i in [50.220, 59.610, 72.650, 61.650, 57.170]]
MRI_t1ce_flair_original = [i/100 for i in [62.280, 61.120, 74.170, 66.300, 54.980]]
df_mri_t1ce_flair_original = get_datafram_MRI(MRI_t1ce_flair_original, "MRI - T1ce\FLAIR Original ", metric)
df_mri_t1ce_original = get_datafram_MRI(MRI_t1ce_original,"MRI - T1ce Original ",metric)
df_mri_t1ce_flair = get_datafram_MRI(MRI_t1ce_flair, "MRI - T1ce\FLAIR Segment. ", metric)
df_mri_t1ce = get_datafram_MRI(MRI_t1ce,"MRI - T1ce Segment. ",metric)
df_HE_1 = get_dataframe(path=r"28-Partial_dataset_MIL_bin_multiclass_kfold\results",
mix="global", loss="CE",name="H&E - global",metric=metric)
df_HE_2 = get_dataframe(path=r"28-Partial_dataset_MIL_bin_multiclass_kfold\results",
mix="mix", loss="CE",name="H&E - mix",metric=metric)
df_HE_3 = get_dataframe(path=r"28-Partial_dataset_MIL_bin_multiclass_kfold\results",
mix="expected", loss="CE",name="H&E - expected",metric=metric)
# df_multi_1 = get_dataframe(path=r"7-Partial_dataset_MIL_bin_MRI_multi_t1ce_flair_cascade_concat_1024\results", mix="mix",
# loss="ordinal",name="H&E + T1ce\FLAIR - Cascade - OE",metric=metric)
df_multi_2 = get_dataframe(path=r"34-Partial_dataset_MIL_bin_MRI_multi_t1ce_flair_cascade_concat_reg_CE\results", mix="expected",
loss="CE",name="H&E + T1ce/FLAIR - Cascade ",metric=metric)
# df_multi_3 = get_dataframe(path=r"9-Partial_dataset_MIL_bin_MRI_multi_t1ce_flair_fusion\results", mix="mix",
# loss="ordinal",name="H&E + T1ce/FLAIR - Mid Fusion - OE",metric=metric)
df_multi_4 = get_dataframe(path=r"31-Partial_dataset_MIL_bin_MRI_multi_t1ce_flair_fusion_CE\results", mix="expected",
loss="CE",name="H&E + T1ce/FLAIR - Mid Fusion ",metric=metric)
df_multi_5 = get_dataframe(path=r"35-Partial_dataset_MIL_bin_MRI_multi_t1ce_flair_fusion_CE_level\results", mix="expected",
loss="CE",name="H&E + T1ce/Flair - Mid Context Fusion ",metric=metric)
df_multi_55 = get_dataframe(path=r"36-Partial_dataset_MIL_bin_MRI_multi_t1ce_fusion_CE_level\results", mix="expected",
loss="CE",name="H&E + T1ce - Mid Context Fusion ",metric=metric)
# df_multi_5 = get_dataframe(path=r"21-Partial_dataset_MIL_bin_MRI_multi_t1ce_flair_fusion_late_ordinal\results", mix="mix",
# loss="ordinal",name="H&E + T1ce/FLAIR - Late Fusion - OE",metric=metric)
df_multi_6 = get_dataframe(path=r"29-Partial_dataset_MIL_bin_MRI_multi_t1ce_flair_late_fusion_CE\results", mix="expected",
loss="CE",name="H&E + T1ce/FLAIR - Late Fusion ",metric=metric)
# df_multi_7 = get_dataframe(path=r"8-Partial_dataset_MIL_bin_MRI_multi_t1ce_cascade_concat_1024\results", mix="mix",
# loss="ordinal",name="H&E + T1ce - Cascade - OE",metric=metric)
df_multi_8 = get_dataframe(path=r"33-Partial_dataset_MIL_bin_MRI_multi_t1ce_cascade_concat_reg_CE\results", mix="expected",
loss="CE",name="H&E + T1ce - Cascade ",metric=metric)
# df_multi_9 = get_dataframe(path=r"24-Partial_dataset_MIL_bin_MRI_multi_t1ce_fusion_ordinal\results", mix="mix",
# loss="ordinal",name="H&E + T1ce - Mid Fusion - OE",metric=metric)
df_multi_10 = get_dataframe(path=r"32-Partial_dataset_MIL_bin_MRI_multi_t1ce_fusion_CE\results", mix="expected",
loss="CE",name="H&E + T1ce - Mid Fusion ",metric=metric)
# df_multi_11 = get_dataframe(path=r"18-Partial_dataset_MIL_bin_MRI_multi_t1ce_fusion_late_ordinal\results", mix="mix",
# loss="ordinal",name="H&E + T1ce - Late Fusion - OE",metric=metric)
df_multi_12 = get_dataframe(path=r"30-Partial_dataset_MIL_bin_MRI_multi_t1ce_late_fusion_CE\results", mix="expected",
loss="CE",name="H&E + T1ce - Late Fusion ",metric=metric)
df_all = pd.concat((df_HE_1,
df_HE_2,
df_HE_3,
# df_HE_4,
# df_HE_5,
# df_HE_6,
df_mri_t1ce_flair_original,
df_mri_t1ce_original,
df_mri_t1ce_flair,
df_mri_t1ce,
# df_multi_1,
df_multi_8,
# df_multi_9,
df_multi_10,
# df_multi_11,
df_multi_12,
df_multi_55,
df_multi_2,
# df_multi_3,
df_multi_4,
df_multi_6,
df_multi_5
# df_multi_7
),axis=0)
return df_all
df_all_auc = get_full_datframe()
df_all_acc = get_full_datframe("Accuracy (%)")
df_all_f1 = get_full_datframe("F1 (%)")
# sns.set(style="darkgrid")
sns.set_theme(style="ticks")
# https://medium.com/swlh/how-to-create-a-seaborn-palette-that-highlights-maximum-value-f614aecd706b
colors= [plt.cm.tab10(i) for i in range(30)]
palette = np.vstack(([colors[6] for i in range(3)],[colors[7] for i in range(4)],[colors[0] for i in range(8)]))
a4_dims = (10, 4)
fig, (ax1, ax2, ax3) = plt.subplots(1,3, figsize=a4_dims) # choose appropriate size to fit your needs
box_bigrams_AUC=sns.despine(fig=None, ax=None, top=True, right=True, left=False, bottom=False, offset=None, trim=False)
box_bigrams_AUC=sns.boxplot(x="AUC (%)", y='Model', data=df_all_auc, palette=palette, showfliers = True, ax=ax1)
box_bigrams_ACC=sns.boxplot(x="Accuracy (%)", y='Model', data=df_all_acc, palette=palette, showfliers = False, ax=ax2)
box_bigrams_F1=sns.boxplot(x="F1 (%)", y='Model', data=df_all_f1, palette=palette, showfliers = False, ax=ax3)
ax2.get_yaxis().set_visible(False)
ax3.get_yaxis().set_visible(False)
ax1.xaxis.grid(True)
ax2.xaxis.grid(True)
ax3.xaxis.grid(True)
box_bigrams_ACC.set_xlabel(box_bigrams_ACC.get_xlabel(),fontsize=14, fontweight='bold')
box_bigrams_F1.set_xlabel(box_bigrams_F1.get_xlabel(),fontsize=14, fontweight='bold')
box_bigrams_AUC.set_xlabel(box_bigrams_AUC.get_xlabel(),fontsize=14, fontweight='bold')
box_bigrams_AUC.set_ylabel(box_bigrams_AUC.get_ylabel(),fontsize=14, fontweight='bold')
fig.savefig('box_plot_all.png', bbox_inches='tight', dpi=200)
# a4_dims = (3, 7)
# fig, ax = plt.subplots(figsize=a4_dims)
# # ax.annotate('global', xy=(0.05, -0.1), xytext=(0.05, -0.2),
# # fontsize=14, ha='center', va='bottom', xycoords='axes fraction',
# # bbox=dict(boxstyle='square', fc='0.8'),
# # arrowprops=dict(arrowstyle='-[, widthB=2.0, lengthB=.5', lw=2.0))
# ax.xaxis.grid(True)
# sns.despine(fig=None, ax=None, top=True, right=True, left=False, bottom=False, offset=None, trim=False)
# box_bigrams = sns.boxplot(ax=ax, x="AUC (%)", y='Model', data=df_all_auc, palette=palette, showfliers = False)
# # box_bigrams.set_xticklabels(box_bigrams.get_xticklabels(),rotation=90)
# box_bigrams.set_xlabel(box_bigrams.get_xlabel(),fontsize=14, fontweight='bold')
# box_bigrams.set_ylabel(box_bigrams.get_ylabel(),fontsize=14, fontweight='bold')
# ax.xaxis.grid(True)
# fig_bigrams = box_bigrams.get_figure()
# fig_bigrams.savefig('boxplot_bigrams.png')