-
Notifications
You must be signed in to change notification settings - Fork 19
/
utils.py
267 lines (204 loc) · 11.1 KB
/
utils.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
import pandas as pd
import numpy as np
from sklearn.preprocessing import StandardScaler
import pickle
import copy
import os
def sum_country_regions(df, country):
# Get indices of the countries
country_indices = df.loc[df['Country/Region'] == country].index
columns = df.columns
has_colony = pd.isna(df.loc[country_indices]).any(axis=1)
if has_colony.any():
# Case 1: imperialist nations
df.drop(has_colony[-has_colony].index, inplace=True)
return df
else:
# Case 2: federal country
# Get the summation data of the countries
# and replace the first region with the result
df.loc[country_indices[0], 4:] = df.loc[country_indices].sum(axis=0)[4:]
# Remove the rest regions
df.drop(country_indices[1:], inplace=True)
return df
def get_change_rates(series, days):
series_change_rate = copy.deepcopy(series)
for day_index in range(len(series)):
min_day = day_index - days if day_index - days >= 0 else 0
change_rate = 1 + (series[day_index] - series[min_day]) / series[min_day] if int(series[min_day]) is not 0 else 1
series_change_rate[day_index] = change_rate
if change_rate >99:
print(change_rate)
return series_change_rate
def generate_df_change_rate(df, days):
df_change_rate = df.copy(deep=True)
columns = df.columns
for country in range(len(df)):
df_change_rate.loc[country, columns[4:]] = get_change_rates(df.loc[country, columns[4:]], days)# .apply(lambda x: get_change_rates(x, days))
return df_change_rate
def sliding_window(data, seq_length, normalize=True):
x=[]
y=[]
for i in range(len(data) - seq_length):
x_ = data[i:(i + seq_length)]
y_ = data[i+seq_length]
if normalize:
x_ = [float(_)/y_ for _ in x_]
y_ = y_/float(y_) if y_ is not 0 else 0
x.append(x_)
y.append(y_)
return x, y
def generate_sequence(df_confirm, df_death, selected_index=None, selected_country=None, normalize=False, seq_length=14):
# Check whether two data have the same number of countries
assert len(df_confirm) == len(df_death)
seq_confirm_x, seq_confirm_y = [], []
seq_death_x, seq_death_y = [], []
# Generate sequence data
# the actual data starts from column index 4
for country in range(len(df_confirm)):
try: df_confirm.loc[country, 'Country/Region']
except Exception as e: continue
if df_confirm.loc[country, 'Country/Region'] not in selected_country:
continue
seq_confirm_x_tmp, seq_confirm_y_tmp = sliding_window(df_confirm.loc[country][selected_index[country]:], seq_length, normalize)
seq_death_x_tmp, seq_death_y_tmp = sliding_window(df_death.loc[country][selected_index[country]:], seq_length, normalize)
seq_confirm_x.extend(seq_confirm_x_tmp)
seq_confirm_y.extend(seq_confirm_y_tmp)
seq_death_x.extend(seq_death_x_tmp)
seq_death_y.extend(seq_death_y_tmp)
seq_death_x = np.expand_dims(np.array(seq_death_x, dtype=np.float), axis=2)
seq_death_y = np.expand_dims(np.array(seq_death_y, dtype=np.float), axis=2)
seq_confirm_x = np.expand_dims(np.array(seq_confirm_x, dtype=np.float), axis=2)
seq_confirm_y = np.expand_dims(np.array(seq_confirm_y, dtype=np.float), axis=2)
print(seq_death_x.shape)
print(seq_death_y.shape)
print(seq_confirm_x.shape)
print(seq_confirm_y.shape)
# Merge data to make 2-dim sequence data
X = np.concatenate((seq_confirm_x, seq_death_x), 2)
Y = np.concatenate((seq_confirm_y, seq_death_y), 1)
print(X.shape)
print(Y.shape)
return X, Y
def generate_COVID_input(df_death,
df_death_change_1st_order,
df_death_change_2nd_order,
df_confirm,
df_confirm_change_1st_order,
df_confirm_change_2nd_order,
df_incoming,
google_trend1,google_trend2,google_trend3,google_trend4,
countries_Korea_inbound,
seq_length=14,
end_day = None,
is_7days = False,
scaler_list=None):
# Filter out unrelated countries & columns from data
selected_country = np.array(countries_Korea_inbound.loc[countries_Korea_inbound.visit.eq(1), 'Country'])
selected_country = np.delete(selected_country, np.argwhere(selected_country == 'Korea, South'))
columns = df_death.columns
select = []
for country in range(len(df_death)):
select.append(df_death.loc[country, 'Country/Region'] in selected_country) # Indices where the selected countries are
s_ = np.where(google_trend1.columns == df_incoming.loc[0, 'date'])[0][0]
if end_day is None:
e_ = np.where(google_trend1.columns == df_incoming.loc[len(df_incoming)-1, 'date'])[0][0]
else :
e_ = np.where(google_trend1.columns == end_day)[0][0]
start_day = np.where(columns == df_incoming.loc[0, 'date'])[0][0]
if end_day is None:
end_day = np.where(columns == df_incoming.loc[len(df_incoming)-1, 'date'])[0][0]
else :
end_day = np.where(columns == end_day)[0][0]
# print(start_day,end_day)
selected_country = np.array(df_death.loc[select, 'Country/Region'])
d1 = df_death.loc[select, columns[start_day:end_day+1]]
d2 = df_death_change_1st_order.loc[select, columns[start_day:end_day+1]]
d3 = df_death_change_2nd_order.loc[select, columns[start_day:end_day+1]]
d4 = df_confirm.loc[select, columns[start_day:end_day+1]]
d5 = df_confirm_change_1st_order.loc[select, columns[start_day:end_day+1]]
d6 = df_confirm_change_2nd_order.loc[select, columns[start_day:end_day+1]]
d7 = google_trend1.loc[:, google_trend1.columns[s_:e_+1]]
d8 = google_trend2.loc[:, google_trend2.columns[s_:e_+1]]
d9 = google_trend3.loc[:, google_trend3.columns[s_:e_+1]]
d10 = google_trend4.loc[:, google_trend4.columns[s_:e_+1]]
d1 = np.log(d1+1)
d4 = np.log(d4+1)
#scaling
mean, std = scaler_list[0]
d4 = d4.subtract(mean[select], axis='index').divide(std[select], axis='index')
mean, std = scaler_list[1]
d1 = d1.subtract(mean[select], axis='index').divide(std[select], axis='index')
mean, std = scaler_list[2]
d5 = d5.subtract(mean[select], axis='index').divide(std[select], axis='index')
mean, std = scaler_list[3]
d6 = d6.subtract(mean[select], axis='index').divide(std[select], axis='index')
mean, std = scaler_list[4]
d2 = d2.subtract(mean[select], axis='index').divide(std[select], axis='index')
mean, std = scaler_list[5]
d3 = d3.subtract(mean[select], axis='index').divide(std[select], axis='index')
##
# Clean up the refined data
d1.reset_index(inplace=True, drop=True)
d2.reset_index(inplace=True, drop=True)
d3.reset_index(inplace=True, drop=True)
d4.reset_index(inplace=True, drop=True)
d5.reset_index(inplace=True, drop=True)
d6.reset_index(inplace=True, drop=True)
d7.reset_index(inplace=True, drop=True)
d8.reset_index(inplace=True, drop=True)
d9.reset_index(inplace=True, drop=True)
d10.reset_index(inplace=True, drop=True)
columns = d1.columns
# Generate country-wise 6-dim vectors of 14 days
data = []
for day in range(end_day - start_day+1 - seq_length):
country_dict = {}
for idx, country in enumerate(selected_country):
_ = np.concatenate((
np.expand_dims(np.array(d1.loc[idx, columns[day:(day+seq_length)]], dtype=np.float), axis=1),
np.expand_dims(np.array(d2.loc[idx, columns[day:(day+seq_length)]], dtype=np.float), axis=1),
np.expand_dims(np.array(d3.loc[idx, columns[day:(day+seq_length)]], dtype=np.float), axis=1),
np.expand_dims(np.array(d4.loc[idx, columns[day:(day+seq_length)]], dtype=np.float), axis=1),
np.expand_dims(np.array(d5.loc[idx, columns[day:(day+seq_length)]], dtype=np.float), axis=1),
np.expand_dims(np.array(d6.loc[idx, columns[day:(day+seq_length)]], dtype=np.float), axis=1),
np.expand_dims(np.array(d7.loc[idx, columns[day:(day+seq_length)]], dtype=np.float), axis=1),
np.expand_dims(np.array(d8.loc[idx, columns[day:(day+seq_length)]], dtype=np.float), axis=1),
np.expand_dims(np.array(d9.loc[idx, columns[day:(day+seq_length)]], dtype=np.float), axis=1),
np.expand_dims(np.array(d10.loc[idx, columns[day:(day+seq_length)]], dtype=np.float), axis=1)), axis=-1)
country_dict[country] = _
data.append(country_dict)
target_continent = np.array(df_incoming.loc[:, countries_Korea_inbound.continent.unique()])
target_total = np.array(df_incoming.loc[:, 'sum'])
if end_day is not None:
# target_continent : (n,6) -> (N:39, T:14,D:6); target_total : (n,1) -> (N:39, T:14,D:1)
target_continent = [target_continent[day+seq_length:day+seq_length+seq_length//2,:] if is_7days else target_continent[day+seq_length:day+seq_length*2,:] for day in range(end_day - start_day+1 - seq_length)]
target_total = [target_total[day+seq_length:day+seq_length+seq_length//2,] if is_7days else target_total[day+seq_length:day+seq_length*2,] for day in range(end_day - start_day+1 - seq_length)]
print()
target_continent = np.stack(target_continent,axis=0)
target_total = np.stack(target_total,axis=0)
return data, target_continent, target_total
else :
return data, target_continent[-(end_day - start_day+1 - seq_length):], target_total[-(end_day - start_day+1 - seq_length):]
def generate_COVID_aux_input(df_roaming,
df_infection_ratio,
passenger_flights,
df_incoming):
columns = df_roaming.columns
start_day_roaming = np.where(df_roaming.columns == df_incoming.loc[0, 'date'])[0][0] +13
start_day_infection = np.where(df_infection_ratio.columns == df_incoming.loc[0, 'date'])[0][0] +13
start_day_passenger = np.where(passenger_flights.columns == df_incoming.loc[0, 'date'])[0][0] +13
end_day = np.where(columns == '5/5/20')[0][0]
data = []
for day in range(end_day - start_day_roaming + 1):
country_dict = {}
for idx, country in enumerate(df_roaming.Country):
_ = np.concatenate((
np.expand_dims(np.array(df_roaming.loc[idx, df_roaming.columns[start_day_roaming + day]], dtype=np.float), axis=1),
np.expand_dims(np.array(df_infection_ratio.loc[idx, df_infection_ratio.columns[start_day_infection + day]], dtype=np.float), axis=1),
np.expand_dims(np.array(passenger_flights.loc[idx, passenger_flights.columns[start_day_passenger + day]], dtype=np.float), axis=1),
),
axis=-1)
country_dict[country] = _
data.append(country_dict)
return data