Nov-23-2016, 07:25 PM
Hi All,
I am new to python and have a number of csv files and am trying to get the count of people who have died and suffered a fall or died and not suffered a fall. I also am trying to work out the average age people lived with each of the above conditions but am having issues tweaking a function. The function in the code below is def is65, I have underlined the relevant aspects below. The function currently subtracts DOB from current date to work out the age, what I am trying to do is to subtract the DOB from the DOD instead. The DOD is from a csv file named DEATHREG (I have also bolded and underlined), problem is no matter how I try to reference the DOD in def is65 it does not work. Any help will be welcome
I am new to python and have a number of csv files and am trying to get the count of people who have died and suffered a fall or died and not suffered a fall. I also am trying to work out the average age people lived with each of the above conditions but am having issues tweaking a function. The function in the code below is def is65, I have underlined the relevant aspects below. The function currently subtracts DOB from current date to work out the age, what I am trying to do is to subtract the DOB from the DOD instead. The DOD is from a csv file named DEATHREG (I have also bolded and underlined), problem is no matter how I try to reference the DOD in def is65 it does not work. Any help will be welcome
import csv
from datetime import datetime
def get_years(delta_in):
"""Get years from timedelta"""
return delta_in.days / 365.25
class PersonData(object):
def __init__(self, dem_data):
"""Initialize person data with dictionary of demographic data from GP_DEM.csv
:param dem_data: dict
"""
self.data = dict()
# clinical events
self.clin = []
# hospital events
self.hosp = []
self.list_names = {'GP_CLIN': self.clin, 'HOSPDAT': self.hosp}
# initialize data with dictionary provided
self.add_data(dem_data)
# first fall event will be written here
self.first_fall_event = None
def is65(self):
"""Return True if the person is older than 65 years"""
return get_years(cur_datetime - all_persons[patid].data['DOB']) > 65
def get_first_fall_event_date(self):
"""Get datetime of the first fall event if exists"""
if self.first_fall_event:
return self.first_fall_event
fall_events = []
# iterate through all events
for hosp_event in self.hosp:
# check for event code
event_code = hosp_event['EVENT_CODE']
if event_code[:3] in fall_event_codes:
# if fall event, append to the temporary list of fall events
fall_events.append(hosp_event['EVENT_DATE'])
if fall_events: # if found:
# first date of fall events
first_event = sorted(fall_events)[0]
self.first_fall_event = first_event
return first_event
# if not found:
return False
def add_data(self, data_dict, keys_to_add=None, list_name=''):
"""Add data to this object's dataset"""
# check for necessary keys to add
if keys_to_add:
new_data = {key: value for key, value in data_dict.items() if key in keys_to_add}
else:
new_data = data_dict
# converting date to datetime format
for date_code in ['EVENT_DATE', 'DOB', 'DOD']:
if date_code in new_data:
new_data[date_code] = datetime.strptime(new_data[date_code], '%Y-%m-%d %H:%M:%S')
# if data should go to the list:
if list_name:
if list_name in self.list_names:
self.list_names[list_name].append(new_data)
# if data goings to the dictionary
else:
self.data.update(new_data)
def add_data_from_csv(file_name, keys_to_add=None, list_name=''):
"""Add data from certain csv file to all_persons dictionary
:param file_name: str
:param keys_to_add: list
:param list_name: str
"""
global all_persons
with open(file_name) as input_file:
input_csv = csv.DictReader(input_file, delimiter=',')
# iterating through all lines in csv file
for line in input_csv:
patid = get_patid(line['PATID'])
if patid in all_persons:
all_persons[patid].add_data(line, keys_to_add=keys_to_add, list_name=list_name)
else:
# if PATID not found, do nothing
pass
with open('GP_DEM.csv') as input_file:
input_csv = csv.DictReader(input_file, delimiter=',')
# all persons data dictionary {PATID1: PersonData1, ...}
all_persons = {line['PATID']: PersonData(line) for line in input_csv}
add_data_from_csv('DEATHREG.csv', ['DOD', 'CAUSE'])
add_data_from_csv('GP_CLIN.csv', ['GPID', 'EVENT_DATE', 'EVENT_CODE', 'EVENT_DATA'], list_name='GP_CLIN')
add_data_from_csv('HOSPDAT.csv', ['HOSPID', 'EVENT_DATE', 'EVENT_CODE'], list_name='HOSPDAT')
# {PATID: [years before first fall event, years after the first}
patid_years = dict()
for patid in all_persons:
person = all_persons[patid]
# use only persons with ages > 65 years
if not person.is65():
continue
# get first fall event date-time
first_event = person.get_first_fall_event_date()
# if no fall events, use other persons
if not first_event:
continue
# if person died
if 'DOD' in person.data:
years_after = get_years(person.data['DOD'] - first_event)
else: # if not died, use current date
years_after = get_years(cur_datetime - first_event)
# get years before the first fall event
years_before = get_years(first_event - person.data['DOB'])
# {PATID: [years before first fall event, years after the first}
patid_years[patid] = [years_before, years_after]
print('2)a) Counting how many persons:')
no_falls = 0
with_falls = 0
# iterate through all persons
for patid in all_persons:
person = all_persons[patid]
# check if person is 65 years old, if died, and without fall events
if person.is65() and ('DOD' in person.data):
if not person.get_first_fall_event_date():
no_falls += 1
else:
with_falls += 1
print('i. The person is over the age of 65 years old, has died and DID NOT suffer a fall event:')
print(no_falls)
print('ii. The person is over the age of 65 years old, has died and HAS suffer at least one fall event in their life:')
print(with_falls)