forked from Yu-Group/covid19-severity-prediction
-
Notifications
You must be signed in to change notification settings - Fork 0
/
load_data.py
191 lines (162 loc) · 7.29 KB
/
load_data.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
import numpy as np
import scipy as sp
import pandas as pd
import matplotlib.pyplot as plt
import matplotlib.dates as mdates
import matplotlib.cbook as cbook
from functions import merge_data
from os.path import join as oj
import os
from sklearn.model_selection import train_test_split
import re
from functions import preprocess
from functions import load_usafacts_data
def load_county_level(
data_dir='data',
cached_file='df_county_level_cached.pkl',
cached_file_abridged='df_county_level_abridged_cached.csv',
ahrf_data='hrsa/data_AHRF_2018-2019/processed/df_renamed.pkl',
diabetes='diabetes/DiabetesAtlasCountyData.csv',
voting='voting/county_voting_processed.pkl',
icu='medicare/icu_county.csv',
heart_disease_data="cardiovascular_disease/heart_disease_mortality_data.csv",
stroke_data="cardiovascular_disease/stroke_mortality_data.csv",
unacast="unacast/Unacast_Social_Distancing_Latest_Available_03_23.csv"
):
'''
Params
------
data_dir
path to the data directory
cached_file
path to the cached file (within the data directory)
'''
df_covid = load_usafacts_data.load_daily_data(dir_mod=data_dir)
cached_file = oj(data_dir, cached_file)
cached_file_abridged = oj(data_dir, cached_file_abridged)
ahrf_data = oj(data_dir, ahrf_data)
diabetes = oj(data_dir, diabetes)
voting = oj(data_dir, voting)
icu = oj(data_dir, icu)
heart_disease_data = oj(data_dir, heart_disease_data)
stroke_data = oj(data_dir, stroke_data)
unacast = oj(data_dir, unacast)
# look for cached file in data_dir
if os.path.exists(cached_file):
df = pd.read_pickle(cached_file)
df = pd.merge(df, df_covid, on='countyFIPS')
return df.sort_values('tot_deaths', ascending=False)
# otherwise run whole pipeline
print('loading county level data...')
df = merge_data.merge_data(ahrf_data=ahrf_data,
medicare_group="All Beneficiaries",
voting=voting,
icu=icu,
resp_group="Chronic respiratory diseases",
heart_disease_data=heart_disease_data,
stroke_data=stroke_data,
diabetes=diabetes,
unacast=unacast) # also cleans usafacts data
# basic preprocessing
df = df.loc[:, ~df.columns.duplicated()]
df = df.infer_objects()
# add features
df = preprocess.add_features(df)
# write cached file
print('caching to', cached_file)
df.to_pickle(cached_file)
important_vars = important_keys(df)
df[important_vars].to_csv(cached_file_abridged)
# add covid data
df = pd.merge(df, df_covid, on='countyFIPS')
return df.sort_values('tot_deaths', ascending=False)
def load_hospital_level(data_dir='data_hospital_level',
merged_hospital_level_info='processed/04_hospital_level_info_merged_with_website.csv',
fips_info='processed/02_county_FIPS.csv'):
'''
Params
------
data_dir
path to the hospital data directory
'''
merged_hospital_level_info = oj(data_dir, merged_hospital_level_info)
fips_info = oj(data_dir, fips_info)
county_fips = pd.read_csv(fips_info)
county_fips['COUNTY'] = county_fips.apply(lambda x: re.sub('[^a-zA-Z]+', '', x['COUNTY']).lower(), axis=1)
county_to_fips = dict(zip(zip(county_fips['COUNTY'], county_fips['STATE']), county_fips['COUNTYFIPS']))
hospital_level = pd.read_csv(merged_hospital_level_info)
def map_county_to_fips(name, st):
if type(name) is str:
index = name.find(' County, ')
name = name[:index]
name = re.sub('[^a-zA-Z]+', '', name).lower()
if (name, st) in county_to_fips:
return int(county_to_fips[(name, st)])
return np.nan
hospital_level['countyFIPS'] = hospital_level.apply(lambda x: map_county_to_fips(x['County Name_x'], x['State_x']),
axis=1).astype('float')
hospital_level['IsAcademicHospital'] = (pd.isna(hospital_level['TIN'])==False).astype(int)
hospital_level['IsUrbanHospital'] = (hospital_level['Urban or Rural Designation'] == 'Urban').astype(int)
hospital_level['IsAcuteCareHospital'] = (hospital_level['Hospital Type'] == 'Acute Care Hospitals').astype(int)
# rename keys
remap = {
'#ICU_beds': 'ICU Beds in County',
'Total Employees': 'Hospital Employees',
'County Name_x': 'County Name',
'Facility Name_x': 'Facility Name'
}
hospital_level = hospital_level.rename(columns=remap)
return hospital_level
def important_keys(df):
demographics = ['PopulationEstimate2018',
'PopTotalMale2017', 'PopTotalFemale2017', 'FracMale2017',
'PopulationEstimate65+2017',
'PopulationDensityperSqMile2010',
'CensusPopulation2010',
'MedianAge2010',
# 'MedianAge,Male2010', 'MedianAge,Female2010',
]
# hospital vars
hospitals_hrsa = ['#FTEHospitalTotal2017', "TotalM.D.'s,TotNon-FedandFed2017", '#HospParticipatinginNetwork2017']
hospitals_misc = ["#Hospitals", "#ICU_beds"]
hospitals = hospitals_hrsa + hospitals_misc
age_distr = list([k for k in df.keys() if 'pop' in k.lower()
and '2010' in k
and ('popmale' in k.lower() or 'popfmle' in k.lower())])
mortality = [k for k in df.keys() if 'mort' in k.lower()
and '2015-17' in k.lower()]
# comorbidity (simultaneous presence of multiple conditions) vars
comorbidity_hrsa = ['#EligibleforMedicare2018', 'MedicareEnrollment,AgedTot2017', '3-YrDiabetes2015-17']
comorbidity_misc = ["DiabetesPercentage", "HeartDiseaseMortality", "StrokeMortality", "Smokers_Percentage", 'Respiratory Mortality']
comorbidity = comorbidity_hrsa + comorbidity_misc
# political leanings (ratio of democrat : republican votes in 2016 presidential election)
political = ['dem_to_rep_ratio']
social_dist = ['unacast_n_grade', 'unacast_daily_distance_diff']
important_vars = demographics + comorbidity + hospitals + political + age_distr + mortality + social_dist
return important_vars
def split_data_by_county(df):
np.random.seed(42)
countyFIPS = df.countyFIPS.values
fips_train, fips_test = train_test_split(countyFIPS, test_size=0.25, random_state=42)
df_train = df[df.countyFIPS.isin(fips_train)]
df_test = df[df.countyFIPS.isin(fips_test)]
return df_train, df_test
if __name__ == '__main__':
df = load_county_level()
print('loaded succesfully')
print(df.shape)
print('data including',
[k for k in df.keys() if '#Deaths' in k][-1],
[k for k in df.keys() if '#Cases' in k][-1])
def city_to_countFIPS_dict(df):
'''
'''
# city to countyFIPS dict
r = df[['countyFIPS', 'City']]
dr = {}
for i in range(r.shape[0]):
row = r.iloc[i]
if not row['City'] in dr:
dr[row['City']] = row['countyFIPS']
elif row['City'] in dr and not np.isnan(row['countyFIPS']):
dr[row['City']] = row['countyFIPS']