-
Notifications
You must be signed in to change notification settings - Fork 0
/
data_utils.py
518 lines (444 loc) · 21.1 KB
/
data_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
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
import numpy as np
import pandas as pd
import datetime as dt
import os
import zipfile
from datetime import datetime, timedelta
from urllib.parse import urlparse
study_prefix = "U01"
def get_user_id_from_filename(f):
#Get user id from from file name
temp = f.split(".")
if(len(temp)==6):
return(temp[3])
temp = f.split("/")
if(len(temp)==3):
return(temp[1])
raise ValueError("Can not get user id from file name.")
def get_file_names_from_zip(z, file_type=None, prefix=study_prefix):
#Extact file list
file_list = list(z.filelist)
if(filter is None):
filtered = [f.filename for f in file_list if (prefix in f.filename) and (".csv" in f.filename)]
else:
filtered = [f.filename for f in file_list if (file_type in f.filename and prefix in f.filename)]
return(filtered)
def get_data_catalog(catalog_file, data_file, data_dir, dict_dir):
dc=pd.read_csv(catalog_file)
dc=dc.set_index("Data Product Name")
dc.data_file=data_dir+data_file #add data zip file field
dc.data_dir=data_dir #add data zip file field
dc.dict_dir=dict_dir #add data distionary directory field
return(dc)
def get_data_dictionary(data_catalog, data_product_name):
dictionary_file = data_catalog.dict_dir + data_catalog.loc[data_product_name]["Data Dictionary File Name"]
dd=pd.read_csv(dictionary_file)
dd=dd.set_index("ElementName")
dd.data_file_name = data_catalog.loc[data_product_name]["Data File Name"] #add data file name pattern field
dd.name = data_product_name #add data product name field
dd.index_fields = data_catalog.loc[data_product_name]["Index Fields"] #add index fields
dd.description = data_catalog.loc[data_product_name]["Data Product Description"]
return(dd)
def get_df_from_zip(file_type,zip_file, participants):
#Get participant list from participants data frame
participant_list = list(participants["Participant ID"])
#Open data zip file
z = zipfile.ZipFile(zip_file)
#Get list of files of specified type
file_list = get_file_names_from_zip(z, file_type=file_type)
#Open file inside zip
dfs=[]
for file_name in file_list:
sid = get_user_id_from_filename(file_name)
if(sid in participant_list):
f = z.open(file_name)
file_size = z.getinfo(file_name).file_size
if file_size > 0:
df = pd.read_csv(f, low_memory=False)
df["Subject ID"] = sid
dfs.append(df)
else:
print('warning %s is empty (size = 0)' % file_name)
df = pd.concat(dfs)
return(df)
def fix_df_column_types(df, dd):
#Set Boolean/String fields to string type to prevent
#interpretation as numeric for now. Leave nans in to
#indicate missing data.
for field in list(df.keys()):
if not (field in dd.index): continue
dd_type = dd.loc[field]["DataType"]
if dd_type in ["Boolean","String","Categorical"]:
if field == 'url':
urls = df[field].values
for index, url in enumerate(urls):
parsed = urlparse(url)
df[field].values[index] = parsed.path[1:]
else:
df[field] = df[field].map(lambda x: x if str(x).lower()=="nan" else str(x))
elif dd_type in ["Ordinal"]:
df[field] = df[field].map(lambda x: x if str(x).lower()=="nan" else int(x))
elif dd_type in ["Time"]:
df[field] = df[field].map(lambda x: x if str(x).lower()=="nan" else pd.to_timedelta(x))
elif dd_type in ["Date"]:
df[field] = df[field].map(lambda x: x if str(x).lower()=="nan" else datetime.strptime(x, "%Y-%m-%d"))
elif dd_type in ["DateTime"]:
#Keep only time for now
max_length = max([len(str(x).split(':')[-1]) for x in df[field].values]) # length of last item after ':'
if max_length < 6: # this includes time with AM/PM
df[field] = df[field].map(lambda x: x if str(x).lower()=="nan" else pd.to_timedelta(x[11:]))
else: # for example: 2020-06-12 23:00:1592002802
df[field] = df[field].map(lambda x: x if str(x).lower()=="nan" else
pd.to_timedelta(pd.to_datetime(x[:16]).strftime("%H:%M:%S")))
#print('\n%s nlargest(10) =\n%s' % (field, df[field].value_counts().nlargest(10)))
return(df)
def get_participant_info(data_catalog):
file = data_catalog.data_dir + data_catalog.loc["Participant Information"]["Data File Name"]
df = pd.read_csv(file)
return(df)
def get_participants_by_type(data_catalog, participant_type):
pi = get_participant_info(data_catalog)
check_type = []
for type_i in pi["Participant Type"].values:
if str(type_i).find(participant_type) >= 0:
check_type.append(True)
else:
check_type.append(False)
pi = pi[check_type]
return(pi)
def crop_data(participants_df, df, b_display, b_crop_end=True):
#Crop before the intervention start date
#Set b_crop_end = True to also crop after the end date (for withdrew status)
participants_df = participants_df.set_index("Participant ID")
fields = list(df.keys())
#Create an observation indicator for an observed value in any
#of the above fields. Sort to make sure data frame is in date order
#per participant
obs_df = 0+((0+~df[fields].isnull()).sum(axis=1)>0)
obs_df.sort_index(axis=0, inplace=True,level=1)
#Get the participant ids according to the data frame
participants = list(obs_df.index.levels[0])
frames = []
for p in participants:
intervention_date = participants_df.loc[p]['Intervention Start Date']
dates = pd.to_datetime(obs_df[p].index)
#Check if there is any data for the participant
if(len(obs_df[p]))>0:
new_obs_df = obs_df[p].copy()
if str(intervention_date).lower() != "nan":
#Check if intervention date is past today's date
intervention_date = pd.to_datetime(intervention_date)
new_obs_df = new_obs_df.loc[dates >= intervention_date]
dates = pd.to_datetime(new_obs_df.index)
today = pd.to_datetime(dt.date.today())
if (intervention_date > today) and b_display:
print('{:<3} intervention date {} is past today\'s date {}'.format(
p, intervention_date.strftime('%Y-%m-%d'), today.strftime('%Y-%m-%d')))
#Crop before the intervention start date
dates_df = pd.to_datetime(df.loc[p].index)
new_df = df.loc[p].copy()
new_df = new_df.loc[dates_df >= intervention_date]
if b_crop_end:
status = participants_df.loc[p]["Participant Status"]
end_date = participants_df.loc[p]['End Date']
if status == 'withdrew':
end_date = pd.to_datetime(end_date)
dates_df = pd.to_datetime(new_df.index)
new_df = new_df.loc[dates_df <= end_date]
new_df['Subject ID'] = p
new_df = new_df.reset_index()
date_name = 'DATE'
columns = list(new_df.columns)
if date_name not in columns:
for col_name in columns:
if col_name.find('Date') >= 0:
date_name = col_name
new_df['Date'] = pd.to_datetime(new_df[date_name]).dt.strftime('%Y-%m-%d')
new_df = new_df.set_index(['Subject ID', 'Date'])
frames.append(new_df)
else:
if b_display:
status = participants_df.loc[p]["Participant Status"]
if (status != 'withdrew') and (str(status).lower() != 'nan'):
print('{:<3} ({}) missing intervention start date'.format(p, status))
continue
if len(frames) > 0:
df = pd.concat(frames)
df = df.sort_index(level=0)
return df
def crop_end_fitbit_per_minute(data_product, participants_df, df, b_display):
#For Fitbit Data Per Minute, we only crop after the end date (for withdrew status)
#Fitbit Data Per Minute has 'Subject ID', 'time' as indices and 'date' as column
participants_df = participants_df.set_index("Participant ID")
fields = list(df.keys())
initial_indices = df.index.names
#Create an observation indicator for an observed value in any
#of the above fields. Sort to make sure data frame is in date order
#per participant
obs_df = 0+((0+~df[fields].isnull()).sum(axis=1)>0)
obs_df.sort_index(axis=0, inplace=True,level=1)
#Get the participant ids according to the data frame
participants = list(obs_df.index.levels[0])
frames = []
for p in participants:
#Check if there is any data for the participant
if(len(obs_df[p]))>0:
new_df = df.loc[p].copy()
date_name = ''
if 'Date' in new_df:
date_name = 'Date'
elif 'date' in new_df:
date_name = 'date'
if date_name != '':
status = participants_df.loc[p]["Participant Status"]
if status == 'withdrew':
new_df[date_name] = pd.to_datetime(new_df[date_name])
end_date = pd.to_datetime(participants_df.loc[p]['End Date'])
new_df = new_df.loc[new_df[date_name] <= end_date]
new_df[date_name] = new_df[date_name].dt.strftime('%Y-%m-%d')
if (b_display):
print('%s: cropped after %s for withdrew participant %s' % (
data_product, end_date.strftime('%Y-%m-%d'), str(p)))
new_df['Subject ID'] = p
new_df = new_df.reset_index()
new_df = new_df.set_index(initial_indices)
frames.append(new_df)
if len(frames) > 0:
df = pd.concat(frames)
df = df.sort_index(level=0)
if (b_display):
print('\nchecking data types...\n')
return df
def load_data(data_catalog, data_product, b_crop=True, b_display=True):
participant_df = get_participants_by_type(data_catalog,"full")
data_dictionary = get_data_dictionary(data_catalog, data_product)
df = get_df_from_zip(data_dictionary.data_file_name, data_catalog.data_file, participant_df)
index = [x.strip() for x in data_dictionary.index_fields.split(";")]
df = df.set_index(index)
df = df.sort_index(level=0)
if (b_crop) and (data_product != 'Fitbit Data Per Minute'):
df = crop_data(participant_df, df, b_display, b_crop_end=True)
elif (b_crop) and (data_product == 'Fitbit Data Per Minute'):
df = crop_end_fitbit_per_minute(data_product, participant_df, df, b_display)
df = fix_df_column_types(df,data_dictionary)
df.name = data_dictionary.name
return(df)
def load_baseline(data_catalog, data_product, filename):
data_dictionary = get_data_dictionary(data_catalog, data_product)
df = pd.read_csv(filename)
index = [x.strip() for x in data_dictionary.index_fields.split(";")]
df = df.set_index(index)
df = fix_df_column_types(df,data_dictionary)
df.sort_index(level=0)
df.name = data_dictionary.name
return(df)
def get_subject_ids(df, b_isbaseline=False):
if b_isbaseline:
sids = df.index.astype(str)
else:
sids = list(df.index.levels[0])
return list(sids)
def get_variables(df):
numerical_types = [np.dtype('int64'), np.dtype('float64')]
cols = [c for c in list(df.columns) if df.dtypes[c] in numerical_types]
return(cols)
def get_catalogs(catalog_file):
df = pd.read_csv(catalog_file)
df = df["Data Product Name"]
df = df[df.values != "Participant Information"]
df = df[df.values != "Baseline Survey"]
return list(df)
def get_categories(dd, field):
categories = dd.loc[field]['Notes'].split(' | ')
return categories
def resample_fitbit_per_minute(participant='105', df=None, filename=None, interval='30Min', b_dropna=True):
#1. Set df to desired input df, or set filename to load df (df=None)
#2. Set participant ID, for example: '105'
#3. Set interval for resampling, for example: '30Min'
if filename != None:
print('loading data for participant %s from %s' % (participant, filename))
df = pd.read_csv(filename, low_memory=False)
else:
print('getting data for participant', participant)
df = df.reset_index()
df = df.groupby(by='Subject ID').get_group(participant)
#Temporary fix for data export issue: replace S with 00 in time format
df['time'] = df['time'].map(lambda x: str(x).replace('S', '00'))
df['datetime'] = df['date'].astype(str) + ' ' + df['time'].astype(str)
df['datetime'] = pd.to_datetime(df['datetime'], format='%Y-%m-%d %H:%M:%S')
df = df.set_index('datetime')
df = df.resample(interval, level=0).first()
df = df.reset_index().set_index(['Subject ID', 'time'])
df.sort_index(level=0)
df.name = 'Fitbit Data Per Minute'
if b_dropna:
df = df.dropna()
return df
def merge_data_frames(dc, data_set_names, short_names):
dfs={}
dds={}
columns={}
#Get all dataframes and columns
for name in data_set_names:
df = load_data(dc, name, b_crop=True, b_display=True)
dfs[name] = df
columns[name] = list(df.columns)
dds[name] = get_data_dictionary(dc,name)
#Merge data sets
for name in data_set_names:
#Get column names in other frames
all_cols = []
for key in columns:
if(key != name):
all_cols = all_cols + columns[key]
all_cols = set(all_cols)
#Get overlapping column names
overlap_cols = list(set(columns[name]).intersection(all_cols))
#Re-map column names
name_map = {x: short_names[name] + " " + x for x in overlap_cols}
dfs[name] = dfs[name].rename(name_map,axis=1,errors='raise')
dds[name] = dds[name].rename(name_map,axis=0,errors='raise')
#Concatenate frames with re-mapped names
df = pd.concat([dfs[name] for name in data_set_names],axis=1)
dd = pd.concat([dds[name] for name in data_set_names],axis=0)
#dd = dd.drop(labels=["Date","Subject ID"])
return(df,dd)
def relabel_participants(df,id_map=None):
#Do a basic relabeling of participant IDs
if id_map is None:
if isinstance(df.index, pd.MultiIndex):
ids = list(df.index.levels[0])
else:
ids = list(df.index)
new_ids = np.random.permutation(len(ids))
id_map_list = list(zip(ids,new_ids ))
id_map = {x[0]:x[1] for x in id_map_list }
print("Creating new ID map")
else:
print("Using supplied ID map")
#Drop any participants not in the mapping
if isinstance(df.index, pd.MultiIndex):
participants_in_index = list(df.index.levels[0])
else:
participants_in_index = list(df.index)
participants_in_index = [str(x) for x in participants_in_index]
participants_in_map = [str(x) for x in id_map.keys()]
participants_to_drop = list(set(participants_in_index) - set(participants_in_map))
df = df.drop(labels = participants_to_drop )
#Re-map the ids
df = df.rename(id_map)
return(df,id_map)
def add_date_indicators(df,dd):
#add collection of date related indicators
dates = pd.to_datetime(df.index.get_level_values(1))
#Day of week
df["Day of Week"] = dates.dayofweek
dd = dd.append(pd.Series({"DataType": "Integer", "Required": "True", "ElementDescription":"Day of the week. 0 is Monday. 6 is Sunday", "ValueRange": "{0..6}"},name="Day of Week"))
#Is weekend day
df["Is Weekend Day"] = dates.dayofweek >=5
dd = dd.append(pd.Series({"DataType": "Boolean", "Required": "True", "ElementDescription":"Is this day a weekend day. True or False.", "ValueRange": "{0,1}"},name="Is Weekend Day"))
#Day of year
df["Day of Year"] = dates.dayofyear
dd = dd.append(pd.Series({"DataType": "Integer", "Required": "True", "ElementDescription":"Day of the year. Jan 1 is day 1.", "ValueRange": "{1..365}"},name="Day of Year"))
#Get days in study variable for each participant
study_days = []
for id in list(df.index.levels[0]):
dates = pd.to_datetime(df.loc[id].index)
date_diff = dates-dates[0]
study_days = study_days + list(date_diff.days)
df["Study day"] = study_days
dd = dd.append(pd.Series({"DataType": "Integer", "Required": "True", "ElementDescription":"Days since start of study. First day is day 0.", "ValueRange": "{0,...}"},name="Study day"))
return df,dd
def get_fb_df_from_zip(file_type,zip_file, participants,interval=None,crop=True):
#Get participant list from participants data frame
participant_list = list(participants["Participant ID"])
participants["Start Date"] = participants["Intervention Start Date"].apply(pd.to_datetime)
participants["End Date"] = participants["End Date"].apply(pd.to_datetime)
participants = participants.set_index("Participant ID")
#Open data zip file
z = zipfile.ZipFile(zip_file)
#Get list of files of specified type
file_list = get_file_names_from_zip(z, file_type=file_type)
#Open file inside zip
dfs=[]
for count,file_name in enumerate(file_list):
sid = get_user_id_from_filename(file_name)
if(sid not in participant_list):
print("Processing ID %s (%d/%d)"%(sid,count,len(file_list)))
print(" ID not in participants list")
else:
print("Processing ID %s (%d/%d)"%(sid,count,len(file_list)))
f = z.open(file_name)
file_size = z.getinfo(file_name).file_size
if file_size > 0:
df = pd.read_csv(f, low_memory=False)
df["Participant ID"] = sid
df['time'] = df['time'].map(lambda x: str(x).replace('S', '00'))
df['datetime'] = df['date'].astype(str) + ' ' + df['time'].astype(str)
df['datetime'] = pd.to_datetime(df['datetime'], format='%Y-%m-%d %H:%M:%S')
#Require both steps and heart rate to not be nan
#Set both to nan if either is and consider minute to
#be invalid in this case
df['valid_minutes'] = np.logical_and(df["steps"].notna(), df["heart_rate"].notna())
df.loc[df["valid_minutes"]==False, 'steps'] = np.nan
df.loc[df["valid_minutes"]==False, 'heart_rate'] = np.nan
if(interval is not None):
df1 = df.set_index('datetime').resample(interval).first()
df2 = df.set_index('datetime').resample(interval).sum()
df3 = df.set_index('datetime').resample(interval).mean()
df1["steps"] = df2["steps"]
df1["valid_minutes"] = df2["valid_minutes"]
df1["heart_rate"] = df3["heart_rate"]
df1=df1.reset_index()
df1=df1.drop(columns=["username","fitbit_account"])
df = df1
if(crop):
df=df.set_index(["datetime"])
start=participants.loc[sid]["Start Date"]
end=participants.loc[sid]["End Date"]
if(start is pd.NaT):
print(" Participant %s start date is missing"%sid)
if(end is pd.NaT):
print(" Participant %s end date is missing"%sid)
else:
df = df[:end]
else:
if(end is pd.NaT):
print(" Participant %s end date is missing"%sid)
df = df[start:]
else:
print( " cropping dates to ",start, " ", end)
df = df[start:end]
df = df.reset_index()
df=df.set_index(["Participant ID","datetime"])
df.loc[df["valid_minutes"]==0, 'steps'] = np.nan
df.loc[df["valid_minutes"]==0, 'heart_rate'] = np.nan
print(" has %d rows"%len(df))
dfs.append(df)
else:
print('warning %s is empty (size = 0)' % file_name)
df = pd.concat(dfs)
return(df)
def apply_transforms(df,dd,transforms):
for t in transforms:
t_type = t["type"]
#Drop the columns
if(t_type=="drop"):
if(t["col"] in df.columns):
df=df.drop(labels=t["col"],axis=1)
if(t["col"] in list(dd.index)):
dd=dd.drop(labels=t["col"],axis=0)
#Add a missing indicator
elif(t_type=="miss_ind"):
df[t["new_name"]] = df[t["col"]].notna()
dd = dd.append(pd.Series({"DataType": "Boolean", "Required": "True", "ElementDescription":t["desc"], "ValueRange": "{True, False}"},name=t["new_name"]))
#Rename columns:
elif(t_type=="rename"):
df=df.rename({t["col"]:t["new_name"]},axis=1)
dd=dd.rename({t["col"]:t["new_name"]},axis=0)
#Merge columns by averaging
elif(t_type=="avg"):
df[t["new_name"]] = df[t["cols"]].mean(axis=1)
dd = dd.append(pd.Series({"DataType": "Float", "Required": "True", "ElementDescription":t["desc"], "ValueRange": ""},name=t["new_name"]))
print("Producing average column")
return(df,dd)