def test_cum_baseline_hazard(self):
cph = CoxPHSurvivalAnalysis()
cph.fit(self.x.values, self.y)
expected_x = numpy.array([
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, 30, 31, 32, 33, 34, 35, 36, 37,
38, 39, 40, 42, 43, 44, 45, 46, 47, 48, 49, 50, 52
])
assert_array_almost_equal(cph.cum_baseline_hazard_.x, expected_x)
expected_y = numpy.array([
0.00678640369024364, 0.0135929334270716, 0.0204043079886091,
0.0272294776707967, 0.0340761479284598, 0.0409513630548852,
0.0478644598407522, 0.0824874121212009, 0.096533196335404,
0.103576022547612, 0.117797217724239, 0.13229957774496,
0.139567689198792, 0.161464121391667, 0.176251599922103,
0.191124391501441, 0.213510322480229, 0.236290954886606,
0.251649773106939, 0.290325398473108, 0.305965134135433,
0.313818453028679, 0.321716523315376, 0.35335069072333,
0.377266607384033, 0.401708829897942, 0.418104688092493,
0.434591703603588, 0.45124335888492, 0.459626329898386,
0.476473421812951, 0.493441590730406, 0.510649810315838,
0.54536240502959, 0.571724727186497, 0.607219385133454,
0.616212515733231, 0.634272429676232, 0.670563043622984,
0.689028239653618, 0.72608698374096, 0.744888154417096,
0.763829951751727, 0.802133842428817, 0.811813515937835,
0.831261170527727, 0.880363253205648, 0.910240767958261,
0.950727380604515
])
actual_y = [cph.cum_baseline_hazard_(v) for v in expected_x]
# check that values increase
self.assertTrue((numpy.diff(actual_y) > 0).all())
assert_array_almost_equal(actual_y, expected_y)
def test_fit_efron(rossi, coef_rossi_coxph_efron):
cph = CoxPHSurvivalAnalysis(ties="efron")
cph.fit(rossi.x.values, rossi.y)
actual = pandas.Series(cph.coef_, index=rossi.x.columns)
assert_array_almost_equal(coef_rossi_coxph_efron.values,
actual.loc[coef_rossi_coxph_efron.index].values)
def test_fit_breslow(rossi, coef_rossi_coxph_breslow):
cph = CoxPHSurvivalAnalysis()
cph.fit(rossi.x.values, rossi.y)
actual = pandas.Series(cph.coef_, index=rossi.x.columns)
assert_array_almost_equal(coef_rossi_coxph_breslow.values,
actual.loc[coef_rossi_coxph_breslow.index].values)
def test_convergence(rossi):
cph = CoxPHSurvivalAnalysis(n_iter=1)
with pytest.warns(
ConvergenceWarning,
match=
"Optimization did not converge: Maximum number of iterations has been exceeded."
):
cph.fit(rossi.x.values, rossi.y)
def test_alpha(self):
cph = CoxPHSurvivalAnalysis(alpha=-0.0001)
self.assertRaisesRegex(ValueError,
r"alpha must be positive, but was -0\.0001",
cph.fit, self.x.values, self.y)
cph.set_params(alpha=-1.25)
self.assertRaisesRegex(ValueError,
r"alpha must be positive, but was -1\.25",
cph.fit, self.x.values, self.y)
def test_convergence(self):
cph = CoxPHSurvivalAnalysis(n_iter=1)
with warnings.catch_warnings(record=True) as w:
warnings.simplefilter("always")
cph.fit(self.x.values, self.y)
self.assertEqual(1, len(w))
self.assertEqual(
"Optimization did not converge: Maximum number of iterations has been exceeded.",
str(w[0].message))
def test_fit_unpenalized():
X, y = load_breast_cancer()
included = X["grade"] != "unkown"
X = X.loc[included, :]
y = y[included.values]
X["grade"] = pandas.Series(pandas.Categorical(
X["grade"].astype(object),
categories=["intermediate", "poorly differentiated",
"well differentiated"]),
index=X.index, name="grade")
enc = OneHotEncoder()
X = enc.fit_transform(X)
cols_unpen = ['age', 'size', 'grade=poorly differentiated',
'grade=well differentiated', 'er=positive']
X = pandas.concat((
X.loc[:, cols_unpen],
X.drop(cols_unpen, axis=1)),
axis=1)
alphas = numpy.ones(X.shape[1])
alphas[:len(cols_unpen)] = 0.0
cph = CoxPHSurvivalAnalysis(alpha=alphas)
cph.fit(X, y)
coef = numpy.array([
-0.0228825990482334, 0.635554486750423, -0.242079636336473,
-1.30197563647684, -2.27790151300312,
0.291950212930807, 0.210861165049552, -0.612456645638769, -0.453414844486013, -0.1239424190253,
0.196855946938761, 1.08724198521351, -0.313645443818603, -0.660016141198812, 1.07104977404073,
0.559632480471393, -0.47740746012516, -1.26199769642326, -1.40486191330444, -0.418517018253652,
0.284936091689505, -0.215531076378674, -0.200889269720281, 0.341231176941461, 0.0307350667648337,
-0.212527052910377, -0.3019678509188, 0.54491723178866, -0.286914381308269, 0.370374100647823,
-0.496258248067704, 0.624528657777646, 0.287884026214139, 0.022095151910937, 0.910293732936019,
-0.13076488639207, 0.0857209529827562, -0.0922302696963889, 0.498136631416287, 0.937133644376614,
0.395090607856869, -1.04727952099579, -0.54974694800345, 0.442372971174454, -0.745558450753062,
-0.0920496108021893, 0.75549238586293, 0.562496351046743, 0.259183349320614, 0.405816113039412,
-0.0969485695700491, -0.507388915258978, -0.474246597197329, -0.209335517183595, 0.187390427612498,
-0.0522568530719332, 0.0806559868641646, -0.0397654339013217, -0.269582356665396, 0.791793553908743,
0.344208857844796, -0.180165785909583, -0.7927695046551, 0.0311635012097026, -0.579429950080662,
-0.264770995160963, 0.869512689697827, 0.765479119494175, -0.173588059680979, -0.199781736503338,
-0.58712767650975, -0.457389854855, 0.3891865514653, 0.707309743580534, -0.121997864690072,
0.0447174402649954, 0.0319336975869795, 0.0117988435665652, -0.593691059339064, -0.838107176656365,
-0.247955128152877
])
assert_array_almost_equal(cph.coef_, coef)
def test_verbose(self):
cph = CoxPHSurvivalAnalysis(verbose=99)
cph.fit(self.x.values, self.y)
cph.set_params(n_iter=1)
with warnings.catch_warnings(record=True):
warnings.simplefilter("ignore")
cph.fit(self.x.values, self.y)
def test_predict(rossi):
cph = CoxPHSurvivalAnalysis()
xc = standardize(rossi.x, with_std=False)
cph.fit(xc.values, rossi.y)
expected = numpy.array([-0.136002823953217, -1.13104636905577, 0.741965816026403, -0.98072115186145,
-0.600098931134794, -0.997407014712788, -0.0993800739865776, -0.266761246895696,
-0.665145743277517, -0.418747210463951, -0.0770761787926419, 0.411385264707043,
-0.0770761787926419, 0.563114305747799, -1.07096133044073])
idx = numpy.array([15, 77, 79, 90, 113, 122, 134, 172, 213, 219, 257, 313, 364, 395, 409])
pred = cph.predict(xc.iloc[idx, :].values)
assert_array_almost_equal(expected, pred)
def test_alpha(rossi):
cph = CoxPHSurvivalAnalysis(alpha=-0.0001)
with pytest.raises(ValueError, match=r"alpha must be positive, but was -0\.0001"):
cph.fit(rossi.x.values, rossi.y)
cph.set_params(alpha=-1.25)
with pytest.raises(ValueError, match=r"alpha must be positive, but was -1\.25"):
cph.fit(rossi.x.values, rossi.y)
def test_fit(self):
cph = CoxPHSurvivalAnalysis()
cph.fit(self.x.values, self.y)
expected = pandas.Series({
"fin": -0.37902189,
"age": -0.05724593,
"race": 0.31412977,
"wexp": -0.15111460,
"mar": -0.43278257,
"paro": -0.08498284,
"prio": 0.09111154
})
actual = pandas.Series(cph.coef_, index=self.x.columns)
assert_array_almost_equal(expected.values,
actual.loc[expected.index].values)
def test_fit_ridge_2(rossi):
# coxph(Surv(week, arrest) ~ ridge(fin, age, race, wexp, mar, paro, prio,
# theta=19.67, scale=FALSE), data=rossi, ties="breslow")
cph = CoxPHSurvivalAnalysis(alpha=19.67)
cph.fit(rossi.x.values, rossi.y)
expected = pandas.Series({'fin': -0.21145000,
'age': -0.06223214,
'race': 0.11957591,
'wexp': -0.10694088,
'mar': -0.13696844,
'paro': -0.04929119,
'prio': 0.09029133})
actual = pandas.Series(cph.coef_, index=rossi.x.columns)
assert_array_almost_equal(expected.values,
actual.loc[expected.index].values)
def test_fit_ridge_1(rossi):
# coxph(Surv(week, arrest) ~ ridge(fin, age, race, wexp, mar, paro, prio,
# theta=1, scale=FALSE), data=rossi, ties="breslow")
cph = CoxPHSurvivalAnalysis(alpha=1.0)
cph.fit(rossi.x.values, rossi.y)
expected = pandas.Series({'fin': -0.36366779384675196,
'age': -0.057788417088377418,
'race': 0.28960521422300672,
'wexp': -0.15082851149160476,
'mar': -0.3829568076550468,
'paro': -0.08230383874483703,
'prio': 0.090951189830228568})
actual = pandas.Series(cph.coef_, index=rossi.x.columns)
assert_array_almost_equal(expected.values,
actual.loc[expected.index].values)
def test_predict_survival_function(self):
cph = CoxPHSurvivalAnalysis()
xc = standardize(self.x, with_std=False)
cph.fit(xc, self.y)
test_idx = [9, 3, 313, 122, 431]
f = cph.predict_survival_function(xc.values[test_idx, :])
self.assertEqual(len(f), len(test_idx))
expected_x = numpy.array([
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, 30, 31, 32, 33, 34, 35, 36, 37,
38, 39, 40, 42, 43, 44, 45, 46, 47, 48, 49, 50, 52
])
assert_array_almost_equal(f[0].x, expected_x)
expected_y = numpy.array(
[[
0.997461059155262, 0.994921063628358, 0.992385735058868,
0.989851752006058, 0.987316287353143, 0.984776787077271,
0.982229880405155, 0.969572939420418, 0.964484914195169,
0.961943739897691, 0.956832876286024, 0.95164893059497,
0.949061479987368, 0.941308786676055, 0.936108961524705,
0.930908109920408, 0.923134421802298, 0.915290309566414,
0.910039421812024, 0.896950008976804, 0.891710496383746,
0.889091087732452, 0.886464513020254, 0.876021834875649,
0.868208744867335, 0.860295712033317, 0.855028100866882,
0.849763725779484, 0.844479679508505, 0.841831970503126,
0.836536000803172, 0.831235646418842, 0.825894605739379,
0.815224813877833, 0.807213878413131, 0.796552033034344,
0.793873120526445, 0.788520538165122, 0.777873630457932,
0.772511632709885, 0.761861634114672, 0.756514792077836,
0.751165903610347, 0.740464726897091, 0.737784677797559,
0.732429416106706, 0.719080607292328, 0.71107749717937,
0.700374533644241
],
[
0.999043882016474, 0.998085846760273, 0.997128049274331,
0.996169233883962, 0.995208324378805, 0.994244341388665,
0.993275988428092, 0.988440345477267, 0.986485348400118,
0.985506531271212, 0.983533018718637, 0.981524558134704,
0.980519526333744, 0.977497911740073, 0.975462571926908,
0.973419764375874, 0.970353074837982, 0.967242232655452,
0.965150517551612, 0.95990330510734, 0.957789534855612,
0.9567298847967, 0.955665378803038, 0.951413568664925,
0.948211695979523, 0.944950494817014, 0.942769166879369,
0.940580788681338, 0.938375713299261, 0.937267565568928,
0.935044498941264, 0.932810785284511, 0.930550920812338,
0.926008946520639, 0.922574381856697, 0.917970142218167,
0.916807238599598, 0.914476355200083, 0.909810449176314,
0.907445513776142, 0.902717737080621, 0.900328600224742,
0.897927987577297, 0.893093037207259, 0.891875336188489,
0.88943384158804, 0.883299174284255, 0.879587091746084,
0.874581779605126
],
[
0.997047493865252, 0.994094987736893, 0.991149132479536,
0.988206067094847, 0.985262510789073, 0.982315505128104,
0.979361149485527, 0.964697962626958, 0.958812207831876,
0.955874508145704, 0.949969979834156, 0.943986274173002,
0.941001628199154, 0.932066788913148, 0.926080800490561,
0.920099053753847, 0.91116833279606, 0.902169133142995,
0.896152043528463, 0.881177346924239, 0.875193135907483,
0.872203575284039, 0.869207278320275, 0.857309023710775,
0.848421992818626, 0.839434570137547, 0.833459215086135,
0.827493527632619, 0.821511608241697, 0.818516504870906,
0.812530287850531, 0.806545301458209, 0.800520667700624,
0.788504317779599, 0.779499208968453, 0.76753682302423,
0.764535224692751, 0.758542854366408, 0.746643087255005,
0.740660154164664, 0.728796971370325, 0.722851218124105,
0.716910044469078, 0.705044665474413, 0.702077410385152,
0.696153525886713, 0.681418170940623, 0.67260509515153,
0.660844217439567
],
[
0.999277487981792, 0.998553357599042, 0.997829237091254,
0.99710417672379, 0.996377361626995, 0.995648049432879,
0.994915257060425, 0.991253310198176, 0.989771583274112,
0.989029450846354, 0.9875325990774, 0.986008486326716,
0.985245535623559, 0.982950582288741, 0.981403739445045,
0.979850427761197, 0.977517078995485, 0.9751482936019,
0.973554486943764, 0.969552590992207, 0.967938970293575,
0.967129721757568, 0.966316544163349, 0.963066369097215,
0.960616445221447, 0.958119046496724, 0.956447429333005,
0.954769459155805, 0.953077720742534, 0.952227180730954,
0.950520162020759, 0.948803968432749, 0.947066660332597,
0.943571819475399, 0.9409262972535, 0.937376036066338,
0.936478650448705, 0.934679123812276, 0.931073502063006,
0.929244251313332, 0.925583871649869, 0.923732352300608,
0.921870729568298, 0.918117633409758, 0.917171619503635,
0.915273906739705, 0.910499952666305, 0.907607302382901,
0.903702170040254
],
[
0.998608413113071, 0.997214644411471, 0.995821831013303,
0.994428148000897, 0.993032034348724, 0.991632072399662,
0.99022638680066, 0.983216197230745, 0.980386484787682,
0.978970679590968, 0.976118050823751, 0.97321758347997,
0.971767207148281, 0.967410751583306, 0.964479728136077,
0.961540752050541, 0.957133998596443, 0.952670279630427,
0.949672572311715, 0.942165647657932, 0.93914685715923,
0.937634655006455, 0.936116291486253, 0.930059391448493,
0.925506309233532, 0.920876059409491, 0.917783073342697,
0.91468336624063, 0.911563331503877, 0.909996636973231,
0.906856220364748, 0.903704188812084, 0.900518754266508,
0.89412719205641, 0.889303474943224, 0.882849825667986,
0.881222141750833, 0.877962502029932, 0.871448809785976,
0.868153120346739, 0.861576393621591, 0.85825886866239,
0.854929447472668, 0.848236143873358, 0.846553007760733,
0.84318146930209, 0.834728527145727, 0.82962662516592,
0.822762829346384
]])
for i, ff in enumerate(f):
actual_y = [ff(v) for v in expected_x]
# check that values decrease
self.assertTrue((numpy.diff(actual_y) < 0).all())
assert_array_almost_equal(actual_y, expected_y[i, :])
def test_ties(rossi):
cph = CoxPHSurvivalAnalysis(ties="xyz")
with pytest.raises(ValueError, match="ties must be one of 'breslow', 'efron'"):
cph.fit(rossi.x.values, rossi.y)
def test_alpha_array(rossi):
cph = CoxPHSurvivalAnalysis(alpha=numpy.array([], dtype=float))
with pytest.raises(ValueError,
match=r"Length alphas \(0\) must match number of features \(7\)"):
cph.fit(rossi.x.values, rossi.y)
alphas = numpy.ones(rossi.x.shape[1])
alphas[-2] = -1e-4
cph.set_params(alpha=alphas)
with pytest.raises(ValueError, match=r"alpha must be positive, but was"):
cph.fit(rossi.x.values, rossi.y)
cph.set_params(alpha=alphas[:-2])
with pytest.raises(ValueError,
match=r"Length alphas \(5\) must match number of features \(7\)"):
cph.fit(rossi.x.values, rossi.y)
