def EstimateMarriageSurvival(dados, limiar, label=""):
complete = dados[dados >= limiar].dropna()
ongoing = dados[dados < limiar]
hf = survival.EstimateHazardFunction(complete, ongoing, label=label)
sf = hf.MakeSurvival()
return hf, sf
def testSurvival(self):
complete = [1, 2, 3, 4, 5]
ongoing = [3, 4, 5]
hf = survival.EstimateHazardFunction(complete, ongoing)
self.assertAlmostEqual(hf[3], 1/6.0)
self.assertAlmostEqual(hf[5], 0.5)
sf = hf.MakeSurvival()
self.assertAlmostEqual(sf[3], 0.625)
self.assertAlmostEqual(sf[5], 0.234375)
def EstimateSurvival(resp):
"""Estimates the survival curve.
resp: DataFrame of respondents
returns: pair of HazardFunction, SurvivalFunction
"""
complete = resp[resp.notdivorced == 0].duration.dropna()
ongoing = resp[resp.notdivorced == 1].durationsofar.dropna()
hf = survival.EstimateHazardFunction(complete, ongoing)
sf = hf.MakeSurvival()
return hf, sf
def EstimateDivorceSurvival(resp):
"""Estimates the survival curve of marriages that end in divorce.
resp: DataFrame of respondents
return: pair of HazardFunction, SurvivalFunction
"""
# NOTE: filling in missing values would be better than dropping them
complete = resp[resp.notdivorced == 0].duration.dropna()
ongoing = resp[resp.notdivorced == 1].durationsofar.dropna()
hf = survival.EstimateHazardFunction(complete, ongoing)
sf = hf.MakeSurvival()
return hf, sf
def EstimateMarriageSurvival(resp):
"""Estimates the survival curve.
resp: DataFrame of respondents
return: pair of HazardFunction, SurvivalFunction
"""
# NOTE: filling in missing values would be better than dropping them
complete = resp6[resp6.evrmarry == 1].agemarry.dropna()
ongoing = resp6[resp6.evrmarry == 0].age
hf = survival.EstimateHazardFunction(complete, ongoing)
sf = hf.MakeSurvival()
return hf, sf
def SurvivalHaz(introductions, lifetimes):
haz = survival.EstimateHazardFunction(lifetimes, introductions)
sf = haz.MakeSurvival()
# thinkplot.plot(sf,color='Grey')
# plt.xlabel("Age (books)")
# plt.ylabel("Probability of Surviving")
# plt.title('Survial Function')
# thinkplot.show()
# thinkplot.plot(haz,color='Grey')
# plt.title('Hazard Function')
# plt.xlabel("Age (books)")
# plt.ylabel("Percent of Lives That End")
# thinkplot.show()
return sf, haz
def EstSurvival(df):
""" Survival curve estimate
@param: df - dataframe of respondent data
return:
@param: hf - hazard function
@param: sf - hazard function
"""
sad = df[df.notdivorced == 0].duration.dropna()
happy = df[df.notdivorced == 1].duration.dropna() # married
hf = survival.EstimateHazardFunction(sad, happy)
sf = hf.MakeSurvival()
return hf, sf
def EstimateSurvival(resp, cutoff=None):
"""Estimates the survival curve.
resp: DataFrame of respondents
cutoff: where to truncate the estimated functions
returns: pair of HazardFunction, SurvivalFunction
"""
complete = resp.loc[resp.complete, 'complete_var'].dropna()
ongoing = resp.loc[~resp.complete, 'ongoing_var'].dropna()
hf = survival.EstimateHazardFunction(complete, ongoing)
if cutoff:
hf.Truncate(cutoff)
sf = hf.MakeSurvival()
return hf, sf
def SurvivalHaz(introductions, lifetimes, plot=False):
"""Given lists of ages and lifespans, this function calculates the
Hazard and Survial curves. If plot is set True, it will plot them."""
haz = survival.EstimateHazardFunction(lifetimes, introductions)
sf = haz.MakeSurvival()
if plot:
thinkplot.plot(sf, color='Grey')
plt.xlabel("Age (books)")
plt.ylabel("Probability of Surviving")
plt.title('Survial Function')
thinkplot.show()
thinkplot.plot(haz, color='Grey')
plt.title('Hazard Function')
plt.xlabel("Age (books)")
plt.ylabel("Percent of Lives That End")
thinkplot.show()
return sf, haz
def EstimateSurvival(resp, cutoff=None):
"""Estimates the survival curve.
resp: DataFrame of respondents
cutoff: where to truncate the estimated functions
returns: pair of HazardFunction, SurvivalFunction
"""
complete = resp[resp.evrmarry].agemarry_index
ongoing = resp[~resp.evrmarry].age_index
hf = survival.EstimateHazardFunction(complete, ongoing)
if cutoff:
hf.Truncate(cutoff)
sf = hf.MakeSurvival()
return hf, sf
#########################################
## Age at first marriage
#########################################
# clean dataframe and extract sub-groups we need
resp6 = nsfg.ReadFemResp()
resp6.cmmarrhx.replace([9997, 9998, 9999], np.nan, inplace=True)
resp6['agemarry'] = (resp6.cmmarrhx - resp6.cmbirth) / 12.0
resp6['age'] = (resp6.cmintvw - resp6.cmbirth) / 12.0
complete = resp6[resp6.evrmarry == 1].agemarry.dropna()
ongoing = resp6[resp6.evrmarry == 0].age
## estimate hazard function
hf = survival.EstimateHazardFunction(complete, ongoing)
thinkplot.Plot(hf)
thinkplot.Show(xlabel='Age (years)', ylabel='Hazard')
## make survival function from hazard function
sf = hf.MakeSurvival()
thinkplot.Plot(sf)
thinkplot.Show(xlabel='Age (years)', ylabel='Prob unmarried', ylim=[0, 1])
## compute resampling survival with weights
ResampleSurvival(resp6)
## show more data
resp5 = survival.ReadFemResp1995()
resp6 = survival.ReadFemResp2002()
resp7 = survival.ReadFemResp2010()
def EstimateSurvival(resp):
complete = resp[resp.notdivorced == 0].duration
ongoing = resp[resp.notdivorced == 1].durationsofar
hf = survival.EstimateHazardFunction(complete, ongoing)
sf = hf.MakeSurvival()
return hf, sf