def UpdateData(self):
to_long_format(self._df, self.event_name)
self.event = self._df[self.event_name]
self.duration = self._df[self.duration_name]
if np.unique(self.event.to_numpy()).size != 2:
mylog.error('Key \'{}\' must have only two values'.format(
self.event_name))
raise ValueError
new_df = self.df.drop(columns=[self.event_name, self.duration_name],
inplace=False)
self._array = new_df.values
self._case_name = list(new_df.index)
self._feature_name = list(new_df.columns)
def seed_df(self):
df = pd.DataFrame.from_records([{
"id": 1,
"var1": 0.1,
"T": 10,
"E": 1
}, {
"id": 2,
"var1": 0.5,
"T": 12,
"E": 0
}])
return utils.to_long_format(df, "T")
def seed_df(self):
df = pd.DataFrame.from_records([{
'id': 1,
'var1': 0.1,
'T': 10,
'E': 1
}, {
'id': 2,
'var1': 0.5,
'T': 12,
'E': 0
}])
return utils.to_long_format(df, 'T')
def seed_df(self):
df = pd.DataFrame.from_records([
{'id': 1, 'var1': 0.1, 'T': 10, 'E': 1},
{'id': 2, 'var1': 0.5, 'T': 12, 'E': 0}
])
return utils.to_long_format(df, 'T')
df = pd.DataFrame([
[1, 3, True, 1],
[6, 4, False, 0],
[3, 5, True, 1],
[2, 5, False, 1],
[4, 6, True, 1],
[7, 7, True, 0],
[8, 8, False, 0],
[5, 8, False, 1],
[9, 9, True, 0],
[10, 10, True, 0],
], columns=['id', 'time', 'event', 'group'])
df = to_long_format(df, 'time')
cv = pd.DataFrame.from_records([
{"id": 1, "z": 0, "time": 0},
{"id": 6, "z": 1, "time": 0},
{"id": 3, "z": 1, "time": 0},
{"id": 2, "z": 0, "time": 0},
{"id": 4, "z": 0, "time": 0},
{"id": 7, "z": 0, "time": 0},
{"id": 8, "z": 0, "time": 0},
{"id": 5, "z": 0, "time": 0},
{"id": 9, "z": 0, "time": 0},
{"id": 10, "z": 0, "time": 0},
{"id": 1, "z": 0, "time": 3},
{"id": 6, "z": 1, "time": 3},
df = pd.DataFrame([
[1, 3, True, 1],
[6, 4, False, 0],
[3, 5, True, 1],
[2, 5, False, 1],
[4, 6, True, 1],
[7, 7, True, 0],
[8, 8, False, 0],
[5, 8, False, 1],
[9, 9, True, 0],
[10, 10, True, 0],
],
columns=['id', 'time', 'event', 'group'])
df = to_long_format(df, 'time')
cv = pd.DataFrame.from_records([
{
"id": 1,
"z": 0,
"time": 0
},
{
"id": 6,
"z": 1,
"time": 0
},
{
"id": 3,
"z": 1,
[1, 3, True, 1],
[6, 4, False, 0],
[3, 5, True, 1],
[2, 5, False, 1],
[4, 6, True, 1],
[7, 7, True, 0],
[8, 8, False, 0],
[5, 8, False, 1],
[9, 9, True, 0],
[10, 10, True, 0],
],
columns=["id", "time", "event", "group"],
)
df = to_long_format(df, "time")
cv = pd.DataFrame.from_records(
[
{"id": 1, "z": 0, "time": 0},
{"id": 6, "z": 1, "time": 0},
{"id": 3, "z": 1, "time": 0},
{"id": 2, "z": 0, "time": 0},
{"id": 4, "z": 0, "time": 0},
{"id": 7, "z": 0, "time": 0},
{"id": 8, "z": 0, "time": 0},
{"id": 5, "z": 0, "time": 0},
{"id": 9, "z": 0, "time": 0},
{"id": 10, "z": 0, "time": 0},
{"id": 1, "z": 0, "time": 3},
{"id": 6, "z": 1, "time": 3},
# -*- coding: utf-8 -*-
if __name__ == "__main__":
import time
import pandas as pd
from lifelines import CoxTimeVaryingFitter
from lifelines.datasets import load_rossi
from lifelines.utils import to_long_format
df = load_rossi()
df = pd.concat([df] * 20)
df = df.reset_index()
df = to_long_format(df, duration_col="week")
ctv = CoxTimeVaryingFitter()
start_time = time.time()
ctv.fit(df,
id_col="index",
event_col="arrest",
start_col="start",
stop_col="stop")
time_took = time.time() - start_time
print("--- %s seconds ---" % time_took)
ctv.print_summary()
# Merge fund_cv DataFrame
# df = df.merge(fund_cv.drop(columns='name'), on='id', how='inner')
# Reduce to just companies for which we have relevant information
df = df[df.id.isin(fund_cv.id.unique())]
# SMALL SUBSET TEST CODE:
# dr = df[(df.name == 'Twitter') | (df.name == 'Facebook')]
# lr = to_long_format(dr, duration_col='age_at_exit')
# rr = fund_cv.loc[(fund_cv.name == 'Twitter') | (fund_cv.name == 'Facebook'),
# ['id', 'raised_amount_usd', 'time_to_funding']]
# lr = add_covariate_to_timeline(lr, rr, 'id', 'time_to_funding', 'success', cumulative_sum=True)
#------------------------------------------------------------------------------
# Prepare DataFrame for lifelines analysis
#------------------------------------------------------------------------------
lf = to_long_format(df, 'age_at_exit')
# Piped version (possibly faster?)
# lf = df.pipe(to_long_format, 'age_at_exit')\
# .pipe(add_covariate_to_timeline(fund_cv_amt,
# 'id', 'time_to_funding', 'success',
# cumulative_sum=False)\
# NOTE WARNING THIS LINE IS SUPER F*****G SLOW.
# Add raised_amount_usd as time-varying covariate
fund_cv_amt = fund_cv[['id', 'raised_amount_usd', 'time_to_funding']]
lf = add_covariate_to_timeline(lf,
fund_cv_amt,
'id',
'time_to_funding',
'success',
[
[1, 3, True, 1],
[6, 4, False, 0],
[3, 5, True, 1],
[2, 5, False, 1],
[4, 6, True, 1],
[7, 7, True, 0],
[8, 8, False, 0],
[5, 8, False, 1],
[9, 9, True, 0],
[10, 10, True, 0],
],
columns=["id", "time", "event", "group"],
)
df = to_long_format(df, "time")
cv = pd.DataFrame.from_records([
{
"id": 1,
"z": 0,
"time": 0
},
{
"id": 6,
"z": 1,
"time": 0
},
{
"id": 3,
"z": 1,
