forked from CSSEGISandData/COVID-19
/
doc_offsets_corr_china.py
134 lines (111 loc) · 3.92 KB
/
doc_offsets_corr_china.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
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import tools
import logging
import sys
import matplotlib.ticker as mtick
from population_database import population
def build_timeseries(df_base):
df_base.set_index(["Province/State", "Country/Region"])
df_base.drop(["Lat", "Long"], inplace=True, axis=1)
df_base = df_base.groupby(['Country/Region']).sum()
df_base = df_base.transpose()
return df_base
# Load data from CSV
df_confirmed = pd.read_csv('csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_confirmed_global.csv')
df_deaths = pd.read_csv('csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_deaths_global.csv')
df_recovered = pd.read_csv('csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_recovered_global.csv')
df_confirmed = build_timeseries(df_confirmed)
df_deaths = build_timeseries(df_deaths)
df_recovered = build_timeseries(df_recovered)
topCountries = df_confirmed.max().sort_values(ascending=False).head(10).index.tolist()
logging.debug("Countries to be analyzed: %s" % topCountries)
fig, axes = plt.subplots(
nrows=1,
ncols=1,
figsize=(25,10)
)
idx = 0
country = "China"
offset = 0
correlations_deaths = []
correlations_recovered = []
corr_min = 99
corr_max = 0
corr_deaths_max = 0
corr_deaths_max_pos = 0
corr_recovered_max = 0
corr_recovered_max_pos = 0
confirmed = df_confirmed[country].to_list()
deaths = df_deaths[country].to_list()
recovered = df_recovered[country].to_list()
# cut all data with less then 50 confirmed cases
while confirmed[0] < 50:
confirmed = confirmed[1:]
deaths = deaths[1:]
recovered = recovered[1:]
while len(confirmed) > 10: # keep at least 10 days for comparison
df = pd.DataFrame(data={
"confirmed": confirmed,
"deaths": deaths,
"recovered": recovered
})
df.reindex()
corr = df.corr(method="pearson")
corr_deaths = corr.at["confirmed", "deaths"]
corr_recovered = corr.at["confirmed", "recovered"]
if corr_deaths > corr_deaths_max:
corr_deaths_max = corr_deaths
corr_deaths_max_pos = offset
if corr_recovered > corr_recovered_max:
corr_recovered_max = corr_recovered
corr_recovered_max_pos = offset
if corr_deaths > corr_max:
corr_max = corr_deaths
if corr_recovered > corr_max:
corr_max = corr_recovered
if corr_deaths < corr_min:
corr_min = corr_deaths
if corr_recovered < corr_min:
corr_min = corr_recovered
correlations_deaths.append(corr_deaths)
correlations_recovered.append(corr_recovered)
# logging.debug("%s %d %s" % (country, offset, corr))
# logging.debug("correlation confirmed deaths: %f" % corr.at["confirmed", "deaths"])
# logging.debug("correlation confirmed recovered: %f" % corr.at["confirmed", "recovered"])
deaths = deaths[1:]
recovered = recovered[1:]
confirmed = confirmed[:-1]
offset += 1
df = pd.DataFrame(data={
"deaths": correlations_deaths,
"recovered": correlations_recovered
})
ax = df.plot(
title = "Correlations over time (%s)" % country,
lw=3,
ax=axes,
)
ax.set_xlabel("Days offset")
ax.set_ylabel("Correlation")
annot_y_pos = corr_deaths_max - ((corr_deaths_max - corr_min) * 0.13)
ax.annotate('Offset Deaths = %d' % corr_deaths_max_pos,
xy=(corr_deaths_max_pos, corr_deaths_max),
xytext=(corr_deaths_max_pos, annot_y_pos),
arrowprops={
"facecolor": plt.rcParams['axes.prop_cycle'].by_key()['color'][0],
# "shrink": 0.6
},
)
annot_y_pos = corr_recovered_max - ((corr_max - corr_min) * 0.13)
ax.annotate(
'Offset Recovered = %d' % corr_recovered_max_pos,
xy=(corr_recovered_max_pos, corr_recovered_max),
xytext=(corr_recovered_max_pos, annot_y_pos),
arrowprops={
"facecolor": plt.rcParams['axes.prop_cycle'].by_key()['color'][1],
},
)
fig.tight_layout(pad = 3.0)
tools.save_chart_doc(fig,"page_correlations_fig4_corr_china")