def test_match_sas_ci(self, data_set):
sas_ci = -0.195996398, 0.195996398
rd = RiskDifference()
rd.fit(data_set, exposure='exp', outcome='dis')
df = rd.results
npt.assert_allclose(df.loc[df.index == '1'][['RD_LCL', 'RD_UCL']],
[sas_ci])
def test_match_sas_sampledata(self):
sas_rr = -0.045129870
sas_se = 0.042375793
sas_ci = -0.128184899, 0.037925158
df = ze.load_sample_data(False)
rd = RiskDifference()
rd.fit(df, exposure='art', outcome='dead')
npt.assert_allclose(rd.risk_difference[1], sas_rr)
rf = rd.results
npt.assert_allclose(rf.loc[rf.index == '1'][['RD_LCL', 'RD_UCL']], [sas_ci])
npt.assert_allclose(rf.loc[rf.index == '1'][['SD(RD)']], sas_se)
def test_match_inverse_of_risk_difference(self):
df = ze.load_sample_data(False)
rd = RiskDifference()
rd.fit(df, exposure='art', outcome='dead')
nnt = NNT()
nnt.fit(df, exposure='art', outcome='dead')
npt.assert_allclose(nnt.number_needed_to_treat[1], 1/rd.risk_difference[1])
rf = rd.results
nf = nnt.results
npt.assert_allclose(nf.loc[nf.index == '1'][['NNT_LCL', 'NNT_UCL']],
1 / rf.loc[rf.index == '1'][['RD_LCL', 'RD_UCL']])
npt.assert_allclose(nf.loc[nf.index == '1'][['SD(RD)']], rf.loc[rf.index == '1'][['SD(RD)']])
def measures_check():
# 7) Check measures plots
data_set = load_sample_data(False)
rr = RiskRatio()
rr.fit(data_set, exposure='art', outcome='dead')
rr.plot(fmt='*', ecolor='r', barsabove=True, markersize=25)
plt.show()
rd = RiskDifference()
rd.fit(data_set, exposure='art', outcome='dead')
rd.plot()
plt.show()
ord = OddsRatio()
ord.fit(data_set, exposure='art', outcome='dead')
ord.plot()
plt.show()
irr = IncidenceRateRatio()
irr.fit(data_set, exposure='art', outcome='dead', time='t')
irr.plot()
plt.show()
ird = IncidenceRateDifference()
ird.fit(data_set, exposure='art', outcome='dead', time='t')
ird.plot()
plt.show()
def test_multiple_exposures(self, multi_exposures):
rd = RiskDifference()
rd.fit(multi_exposures, exposure='exp', outcome='dis')
assert rd.results.shape[0] == 3
assert list(rd.results.index) == ['Ref:0', '1', '2']
def test_risk_difference_equal_to_0(self, data_set):
rd = RiskDifference()
rd.fit(data_set, exposure='exp', outcome='dis')
assert rd.risk_difference[1] == 0
def test_frechet_bounds2(self, multi_exposures):
rd = RiskDifference()
rd.fit(multi_exposures, exposure='exp', outcome='dis')
npt.assert_allclose(
rd.results['UpperBound'][1:] - rd.results['LowerBound'][1:],
[1.0000, 1.0000])
def test_frechet_bounds(self):
df = ze.load_sample_data(False)
rd = RiskDifference()
rd.fit(df, exposure='art', outcome='dead')
npt.assert_allclose(
rd.results['UpperBound'][1] - rd.results['LowerBound'][1], 1.0000)
def test_match_sas_se(self, data_set):
sas_se = 0.1
rd = RiskDifference()
rd.fit(data_set, exposure='exp', outcome='dis')
df = rd.results
npt.assert_allclose(df.loc[df.index == '1'][['SD(RD)']], sas_se)
#######################################################################################################################
# Binary Outcome
#######################################################################################################################
df = load_sample_data(timevary=False)
df = df.drop(columns=['cd4_wk45'])
df[['cd4_rs1', 'cd4_rs2']] = spline(df, 'cd40', n_knots=3, term=2, restricted=True)
df[['age_rs1', 'age_rs2']] = spline(df, 'age0', n_knots=3, term=2, restricted=True)
#############################
# Naive Risk Difference
from zepid import RiskDifference
rd = RiskDifference()
rd.fit(df, exposure='art', outcome='dead')
rd.summary()
#############################
# G-formula
from zepid.causal.gformula import TimeFixedGFormula
g = TimeFixedGFormula(df, exposure='art', outcome='dead')
g.outcome_model(model='art + male + age0 + age_rs1 + age_rs2 + cd40 + cd4_rs1 + cd4_rs2 + dvl0',
print_results=False)
# Estimating marginal effect under treat-all plan
g.fit(treatment='all')
r_all = g.marginal_outcome
# Estimating marginal effect under treat-none plan
g.fit(treatment='none')
r_none = g.marginal_outcome