lifelines 0.27.5__py3-none-any.whl → 0.27.7__py3-none-any.whl

lifelines/fitters/__init__.py

@@ -1965,7 +1965,6 @@ class ParametricRegressionFitter(RegressionFitter):
             raise exceptions.ConvergenceError(
                 dedent(
                     f"""\
-                {minimum_results=}
 
                 Fitting did not converge. Try the following:
 
@@ -1976,6 +1975,8 @@ class ParametricRegressionFitter(RegressionFitter):
                 4. Try using an alternate minimizer: ``fitter._scipy_fit_method = "SLSQP"``.
                 5. Trying adding a small penalizer (or changing it, if already present). Example: `{self._class_name}(penalizer=0.01).fit(...)`.
                 6. Are there any extreme outliers? Try modeling them or dropping them to see if it helps convergence.
+
+                minimum_results={minimum_results}
             """
                 )
             )

lifelines/fitters/mixins.py

@@ -107,6 +107,49 @@ class ProportionalHazardMixin:
 
         for variable in self.params_.index.intersection(columns or self.params_.index):
             minumum_observed_p_value = test_results.summary.loc[variable, "p"].min()
+            
+            # plot is done (regardless of test result) whenever `show_plots = True`
+            if show_plots:
+                axes.append([])
+                print()
+                print("   Bootstrapping lowess lines. May take a moment...")
+                print()
+                from matplotlib import pyplot as plt
+
+                fig = plt.figure()
+
+                # plot variable against all time transformations.
+                for i, (transform_name, transformer) in enumerate(TimeTransformers().iter(["rank", "km"]), start=1):
+                    p_value = test_results.summary.loc[(variable, transform_name), "p"]
+
+                    ax = fig.add_subplot(1, 2, i)
+
+                    y = residuals_and_duration[variable]
+                    tt = transformer(self.durations, self.event_observed, self.weights)[self.event_observed.values]
+
+                    ax.scatter(tt, y, alpha=0.75)
+
+                    y_lowess = lowess(tt.values, y.values)
+                    ax.plot(tt, y_lowess, color="k", alpha=1.0, linewidth=2)
+
+                    # bootstrap some possible other lowess lines. This is an approximation of the 100% confidence intervals
+                    for _ in range(plot_n_bootstraps):
+                        ix = sorted(np.random.choice(n, n))
+                        tt_ = tt.values[ix]
+                        y_lowess = lowess(tt_, y.values[ix])
+                        ax.plot(tt_, y_lowess, color="k", alpha=0.30)
+
+                    best_xlim = ax.get_xlim()
+                    ax.hlines(0, 0, tt.max(), linestyles="dashed", linewidths=1)
+                    ax.set_xlim(best_xlim)
+
+                    ax.set_xlabel("%s-transformed time\n(p=%.4f)" % (transform_name, p_value), fontsize=10)
+                    axes[-1].append(ax)
+
+                fig.suptitle("Scaled Schoenfeld residuals of '%s'" % variable, fontsize=14)
+                plt.tight_layout()
+                plt.subplots_adjust(top=0.90)
+
             if np.round(minumum_observed_p_value, 2) > p_value_threshold:
                 continue
 
@@ -181,48 +224,9 @@ class ProportionalHazardMixin:
                         ),
                         end="\n\n",
                     )
+#################
 
-            if show_plots:
-                axes.append([])
-                print()
-                print("   Bootstrapping lowess lines. May take a moment...")
-                print()
-                from matplotlib import pyplot as plt
-
-                fig = plt.figure()
-
-                # plot variable against all time transformations.
-                for i, (transform_name, transformer) in enumerate(TimeTransformers().iter(["rank", "km"]), start=1):
-                    p_value = test_results.summary.loc[(variable, transform_name), "p"]
-
-                    ax = fig.add_subplot(1, 2, i)
-
-                    y = residuals_and_duration[variable]
-                    tt = transformer(self.durations, self.event_observed, self.weights)[self.event_observed.values]
-
-                    ax.scatter(tt, y, alpha=0.75)
-
-                    y_lowess = lowess(tt.values, y.values)
-                    ax.plot(tt, y_lowess, color="k", alpha=1.0, linewidth=2)
-
-                    # bootstrap some possible other lowess lines. This is an approximation of the 100% confidence intervals
-                    for _ in range(plot_n_bootstraps):
-                        ix = sorted(np.random.choice(n, n))
-                        tt_ = tt.values[ix]
-                        y_lowess = lowess(tt_, y.values[ix])
-                        ax.plot(tt_, y_lowess, color="k", alpha=0.30)
-
-                    best_xlim = ax.get_xlim()
-                    ax.hlines(0, 0, tt.max(), linestyles="dashed", linewidths=1)
-                    ax.set_xlim(best_xlim)
-
-                    ax.set_xlabel("%s-transformed time\n(p=%.4f)" % (transform_name, p_value), fontsize=10)
-                    axes[-1].append(ax)
-
-                fig.suptitle("Scaled Schoenfeld residuals of '%s'" % variable, fontsize=14)
-                plt.tight_layout()
-                plt.subplots_adjust(top=0.90)
-
+            
         if advice and counter > 0:
             print(
                 dedent(

lifelines/metrics.py

@@ -0,0 +1,60 @@
+# -*- coding: utf-8 -*-
+
+import numpy as np
+from lifelines.utils import concordance_index, ConvergenceError
+
+__all__ = ["concordance_index", "uncensored_l2_log_loss", "uncensored_l1_log_loss"]
+
+
+def uncensored_l1_log_loss(event_times, predicted_event_times, event_observed=None):
+    r"""
+    Calculates the l1 log-loss of predicted event times to true event times for *non-censored*
+    individuals only.
+
+    .. math::  1/N \sum_{i} |log(t_i) - log(q_i)|
+
+    Parameters
+    ----------
+      event_times:
+        a (n,) array of observed survival times.
+      predicted_event_times:
+        a (n,) array of predicted survival times.
+      event_observed:
+        a (n,) array of censored flags, 1 if observed,  0 if not. Default None assumes all observed.
+
+    Returns
+    -------
+      l1-log-loss: a scalar
+    """
+    if event_observed is None:
+        event_observed = np.ones_like(event_times, dtype=bool)
+
+    ix = event_observed.astype(bool)
+    return np.abs(np.log(event_times[ix]) - np.log(predicted_event_times[ix])).mean()
+
+
+def uncensored_l2_log_loss(event_times, predicted_event_times, event_observed=None):
+    r"""
+    Calculates the l2 log-loss of predicted event times to true event times for *non-censored*
+    individuals only.
+
+    .. math::  1/N \sum_{i} (log(t_i) - log(q_i))**2
+
+    Parameters
+    ----------
+      event_times:
+        a (n,) array of observed survival times.
+      predicted_event_times:
+        a (n,) array of predicted survival times.
+      event_observed:
+        a (n,) array of censored flags, 1 if observed,  0 if not. Default None assumes all observed.
+
+    Returns
+    -------
+      l2-log-loss: a scalar
+    """
+    if event_observed is None:
+        event_observed = np.ones_like(event_times, dtype=bool)
+
+    ix = event_observed.astype(bool)
+    return np.power(np.log(event_times[ix]) - np.log(predicted_event_times[ix]), 2).mean()

lifelines/version.py

@@ -1,4 +1,4 @@
 # -*- coding: utf-8 -*-
 from __future__ import unicode_literals
 
-__version__ = "0.27.5"
+__version__ = "0.27.7"

{lifelines-0.27.5.dist-info → lifelines-0.27.7.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: lifelines
-Version: 0.27.5
+Version: 0.27.7
 Summary: Survival analysis in Python, including Kaplan Meier, Nelson Aalen and regression
 Home-page: https://github.com/CamDavidsonPilon/lifelines
 Author: Cameron Davidson-Pilon

{lifelines-0.27.5.dist-info → lifelines-0.27.7.dist-info}/RECORD

@@ -1,10 +1,12 @@
-lifelines/__init__.py,sha256=qPF0ennAJGEAK-s63Z0e4DHiWnfRT98slfHjR-tjGw4,2235
+lifelines/__init__.py,sha256=F_sKrawq6L4GwTPgOu4FjoGUKQ2gfelAVIQOW1Ee8Ao,2241
 lifelines/calibration.py,sha256=Luii7bkJ2YB0jpuOYYhI22aUyEc1gLsS10Pno6Sqo98,4113
 lifelines/exceptions.py,sha256=Kf6GN2vB-SHde2mbPomj2PhpnCvCBOSTUZLY1jwOw-U,514
 lifelines/generate_datasets.py,sha256=WsvvrZt0jEYQ7-Fp53vrCq7MzmAM2pPUSoCaiQRwN5g,10155
+lifelines/metrics.py,sha256=1NY5CAwETP7C8lIUjmQz2ye648W64oChs2zdi-rqEzc,1889
 lifelines/plotting.py,sha256=AKRFHz0ye_pu5kE_nod3kM5AhQKL8h1tdrHt77nDDa4,33150
 lifelines/statistics.py,sha256=cOszUYz87elnbTAt6V3fTrHwPjB9HFI1hxjLvKypS6k,35129
-lifelines/version.py,sha256=JBLJotCZSiROHDoXzVOQQ__ruAm8MRSl4IwA_Zhp12M,88
+lifelines/version.py,sha256=5c99jzzH3waN_PF9HgeSaexk7Z6Rucqb2uRdXVlPMt4,88
+lifelines/datasets/ACTG175.csv,sha256=VvujH6DXv7_5ZntxSf2WqXw1LnKqWChxpiqTXoEvDgc,185144
 lifelines/datasets/CuZn-LeftCensoredDataset.csv,sha256=PxTdZcJPPbhtaadpHjhMFVcUxmSn84BuDarujZIJpm4,1996
 lifelines/datasets/__init__.py,sha256=9OIUjJUP6mK4oiFIRwHTh9tI5icDj4z8owUfPPDiY8c,19990
 lifelines/datasets/anderson.csv,sha256=nTAtTK8mf0ymU88nKvO2Fj0WL9SE9o4S0GVujmX8Cl4,580
@@ -35,7 +37,7 @@ lifelines/datasets/rossi.csv,sha256=AhRAAXDgfzAVooXtyiAUysDa6KrBJfy6rWQkkOBfiSw,
 lifelines/datasets/stanford_heart.csv,sha256=HWS9SqJjQ6gDmvxxKCJLR1cOIJ8XKuwTNu4bW8tKWVM,8859
 lifelines/datasets/static_test.csv,sha256=w2PtSkXknCZfciwqcOZGlA8znBO7jTcq_AJ5e6NStAk,101
 lifelines/datasets/waltons_dataset.csv,sha256=Fd4UX6tGYxgGhXtH3T-S81wIGIbVohv5yom4aw0kXL8,2449
-lifelines/fitters/__init__.py,sha256=-4N1nTjafrvjJbzv15FZPSUKGn5C5Ei_QLzJSzDd5as,151111
+lifelines/fitters/__init__.py,sha256=vQ4AtidwM-gDGkGIzwSwRTB1c2LogpjQirKKrn1kQ-I,151127
 lifelines/fitters/aalen_additive_fitter.py,sha256=vRQb38weMcknyxC9bJwiALwCzxmJ5DsEZwHkz2zV93k,21518
 lifelines/fitters/aalen_johansen_fitter.py,sha256=fIsyi_al5X5QhAzQbrkCreYLjlY_4mcuSEhcm6Ul7Oo,12515
 lifelines/fitters/breslow_fleming_harrington_fitter.py,sha256=Te1Y73lIIKhTC6yMADe35RVHI4XOLF17ub-N8oudS4I,4091
@@ -50,7 +52,7 @@ lifelines/fitters/log_logistic_aft_fitter.py,sha256=cw179z0_IqvuWgOORHSZ1lBiidHc
 lifelines/fitters/log_logistic_fitter.py,sha256=iTH97i9TrLp5IVBIZHC8nx5rvSn2-KM-wfv1wR_YSPU,4004
 lifelines/fitters/log_normal_aft_fitter.py,sha256=aOcdMR8T4vhy2BKGebrpEJD_lTZIQQ5VsrnuuKkU0RA,7890
 lifelines/fitters/log_normal_fitter.py,sha256=NLn1DCxJ9WJrVaairJPcOu_lShko_-vwoXw6goRR42w,3557
-lifelines/fitters/mixins.py,sha256=_ht2WvDztBsvgGQ5vmprhfiChEP5DfdwNQfa9JkyxtY,12399
+lifelines/fitters/mixins.py,sha256=6k5-g8cit8ODbU7PbVD9tfYsY0jpde0HID3wJQ5kz1k,12527
 lifelines/fitters/mixture_cure_fitter.py,sha256=UetFlv9EfFYMDt95M2iR354lna5RKeWtO_lkoaMmoZE,5416
 lifelines/fitters/nelson_aalen_fitter.py,sha256=UNlEX5wR6xsUmEmJ2n2MEqblz-KvGmvlh8eGHfuQf6Y,10666
 lifelines/fitters/npmle.py,sha256=HV3yeu1byVv5oBSdv5TuLUg2X5NUxydxj8-h_xYScB0,10143
@@ -66,8 +68,8 @@ lifelines/utils/lowess.py,sha256=MMydVcnbxqIgsiNcIgVUFtlFycD7v3ezwEGpituvBHs,254
 lifelines/utils/printer.py,sha256=-nXxu02gs0kaKfoQQ65sH-I45tGmgoFeOOIUSEc53iE,5861
 lifelines/utils/safe_exp.py,sha256=HCCAkwQTx6G2qRC03v9Q_GWqVj8at1Eac1JVrMgS9hg,4350
 lifelines/utils/sklearn_adapter.py,sha256=S5qotbZ1hf1fhFBsx39Yd_NpA31jB9HhRiLjE8TWlhw,4202
-lifelines-0.27.5.dist-info/LICENSE,sha256=AasDeD139SnTdfXbKgN4BMyMgBlRy9YFs60tNrB4wf0,1079
-lifelines-0.27.5.dist-info/METADATA,sha256=iydwn5g_0UfE4g4thVCJfiVvtAaKyZ-5F1vjyOzE4pk,3297
-lifelines-0.27.5.dist-info/WHEEL,sha256=pkctZYzUS4AYVn6dJ-7367OJZivF2e8RA9b_ZBjif18,92
-lifelines-0.27.5.dist-info/top_level.txt,sha256=3i57Z4mtpc6jWrsW0n-_o9Y7CpzytMTeLMPJBHYAo0o,10
-lifelines-0.27.5.dist-info/RECORD,,
+lifelines-0.27.7.dist-info/LICENSE,sha256=AasDeD139SnTdfXbKgN4BMyMgBlRy9YFs60tNrB4wf0,1079
+lifelines-0.27.7.dist-info/METADATA,sha256=N0RFCC7WLIaqHa6CYMMvvMkYVWE-Er77nqFYI6m3uWI,3297
+lifelines-0.27.7.dist-info/WHEEL,sha256=pkctZYzUS4AYVn6dJ-7367OJZivF2e8RA9b_ZBjif18,92
+lifelines-0.27.7.dist-info/top_level.txt,sha256=3i57Z4mtpc6jWrsW0n-_o9Y7CpzytMTeLMPJBHYAo0o,10
+lifelines-0.27.7.dist-info/RECORD,,