-
Notifications
You must be signed in to change notification settings - Fork 0
/
CSI.py
197 lines (180 loc) · 6.32 KB
/
CSI.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
# 예보, 관측
# A (Yes, Yes)
# B (No, Yes)
# C (Yes, No)
# D (No, No)
# HR(Hiting Rate) 적중률 : (A+D) / (A+B+C+D)
# POD 탐지율 : A / A+B 결과 YES에 대해 예측 YES의 비율 = (YES,YES) / (?,YES)
# FAR 부적중률 : C / A+C 예측 YES에 대해 결과 NO의 비율 = (YES,NO) / (YES,?)
# CSI 성공지수 : A / A+B+C 결과 YES+온다고했는데안옴 中 비가예보되고온날 = (YES,YES) / (?,YES)+(YES,NO)
from cmath import log10, nan
import numpy as np
from matplotlib import pyplot as plt
from netCDF4 import Dataset
import torch
import os
import random
import math
level = [0.1,1,5,10]
# level = np.log(np.array(level)+0.01)
data_dir = 'C:/Users/ks297/Desktop/AllRainDropUnetData/result/numpy'
lst_data = os.listdir(data_dir)
lst_label = [f for f in lst_data if f.startswith('label')]
lst_output = [f for f in lst_data if f.startswith('output')]
lst_label.sort()
lst_output.sort()
def compare(output, label):
CSI=[]
for i in range(len(level)):
forecast = np.power(10,output)-0.01 >= level[i]
observe = np.power(10,label)-0.01 >= level[i]
hit_mask = forecast & observe
total_mask = forecast | observe
hit_num = np.sum(hit_mask)
total_num = np.sum(total_mask)
if total_num==0:
csi = -1
else :
csi = hit_num/total_num
CSI.append(np.round(csi,4))
#print(CSI)
return CSI
def print_csimean():
csilist=[]
for j in range(len(lst_label)):
output = np.load(data_dir+'/'+lst_output[j])
label = np.load(data_dir+'/'+lst_label[j])
csilist += [compare(output,label)]
csimean = [0,0,0,0]
for j in range(4):
sums = 0
count = 0
for k in range(len(csilist)):
if csilist[k][j]!=-1:
sums += csilist[k][j]
count +=1
if count == 0:
csimean[j]=0
else :
csimean[j]=sums/count
print(np.round(csimean,4))
def show_random():
n=5
fig, axes = plt.subplots(nrows=n, ncols=6)
fig.suptitle('Random Display')
cols = ['Predict','Real','CSI 0.1','CSI 1','CSI 5','CSI 10']
for ax, col in zip(axes[0],cols):
ax.set_title(col)
for k in range(n):
i = random.randint(0,len(lst_label))
outputi = np.load(data_dir+'/'+lst_output[i])
plt.subplot(n,6,1+6*k)
plt.imshow(outputi, cmap='jet',vmin=-2,vmax=2)
plt.axis('off')
labeli = np.load(data_dir+'/'+lst_label[i])
plt.subplot(n,6,2+6*k)
plt.imshow(labeli, cmap='jet',vmin=-2,vmax=2)
plt.axis('off')
for j in range(len(level)):
forecast = (np.power(10,outputi)-0.01 >= level[j])*2
observe = (np.power(10,labeli)-0.01 >= level[j])*2
hit_mask = (forecast & observe)/2
total_mask = (forecast | observe)*1
differ_makst = forecast - observe + hit_mask
plt.subplot(n,6,j+6*k+3)
plt.imshow(differ_makst, cmap='jet',vmin=-2,vmax=2)
plt.axis('off')
# unique,counts = np.unique(differ_makst, return_counts=True)
# unique_dict = dict(zip(unique,counts))
# print(unique_dict)
plt.show()
def print_top5():
csilist=[]
for j in range(len(lst_label)):
output = np.load(data_dir+'/'+lst_output[j])
label = np.load(data_dir+'/'+lst_label[j])
a = compare(output,label)
for b in a:
if b==1:
print(j)
csilist.append(a)
arr = torch.tensor(csilist)
arr = arr.transpose(1,0)
arr = arr.tolist()
top = []
top_index = []
for i in arr:
best = []
best_index = []
for j in range(5):
best.append(max(i))
best_index.append(i.index(max(i)))
i.remove(max(i))
top.append(best)
top_index.append(best_index)
print(top,top_index)
fig, axes = plt.subplots(nrows=4, ncols=5)
fig.suptitle('Top5 Display')
cols = ['Top1','Top2','Top3','Top4','Top5']
for ax, col in zip(axes[0],cols):
ax.set_title(col)
rows = ['CSI 0.1','CSI 1','CSI 5','CSI 10']
for bx, row in zip(axes[:,0],rows):
bx.set_ylabel(row)
for i in range(4):
for j in range(5):
plt.subplot(4,5,j+1+5*i)
plt.xlabel(top[i][j])
plt.imshow(np.load(data_dir+'/'+lst_output[top_index[i][j]]),cmap='jet',vmin=-2.5,vmax=2.5)
cx = plt.gca()
cx.axes.xaxis.set_ticks([])
cx.axes.yaxis.set_ticks([])
plt.show()
def print_top5dif():
csilist=[]
for j in range(len(lst_label)):
output = np.load(data_dir+'/'+lst_output[j])
label = np.load(data_dir+'/'+lst_label[j])
csilist += [compare(output,label)]
arr = torch.tensor(csilist)
arr = arr.transpose(1,0)
arr = arr.tolist()
top = []
top_index = []
for i in arr:
best = []
best_index = []
for j in range(5):
best.append(max(i))
best_index.append(i.index(max(i)))
i.remove(max(i))
top.append(best)
top_index.append(best_index)
print(top,top_index)
fig, axes = plt.subplots(nrows=4, ncols=5)
fig.suptitle('Top5 Display')
cols = ['Top1','Top2','Top3','Top4','Top5']
for ax, col in zip(axes[0],cols):
ax.set_title(col)
rows = ['CSI 0.1','CSI 1','CSI 5','CSI 10']
for bx, row in zip(axes[:,0],rows):
bx.set_ylabel(row)
for i in range(4):
for j in range(5):
outputi = np.load(data_dir+'/'+lst_output[top_index[i][j]])
labeli = np.load(data_dir+'/'+lst_label[top_index[i][j]])
forecast = (np.power(10,outputi)-0.01 >= level[i])*2
observe = (np.power(10,labeli)-0.01 >= level[i])*2
hit_mask = (forecast & observe)/2
differ_makst = forecast - observe + hit_mask
plt.subplot(4,5,j+1+5*i)
plt.xlabel(top[i][j])
plt.imshow(differ_makst,cmap='jet',vmin=-2,vmax=2)
cx = plt.gca()
cx.axes.xaxis.set_ticks([])
cx.axes.yaxis.set_ticks([])
plt.show()
# show_random()
# print_csimean()
# print_top5()
print_top5dif()