scikit-survival 0.24.1__cp313-cp313-macosx_11_0_arm64.whl → 0.26.0__cp313-cp313-macosx_11_0_arm64.whl

sksurv/datasets/base.py

@@ -36,10 +36,10 @@ def _get_x_y_survival(dataset, col_event, col_time, val_outcome, competing_risks
         event_type = np.int64 if competing_risks else bool
         y = np.empty(dtype=[(col_event, event_type), (col_time, np.float64)], shape=dataset.shape[0])
         if competing_risks:
-            y[col_event] = dataset[col_event].values
+            y[col_event] = dataset[col_event].to_numpy()
         else:
-            y[col_event] = (dataset[col_event] == val_outcome).values
-        y[col_time] = dataset[col_time].values
+            y[col_event] = (dataset[col_event] == val_outcome).to_numpy()
+        y[col_time] = dataset[col_time].to_numpy()
 
         x_frame = dataset.drop([col_event, col_time], axis=1)
 
@@ -82,18 +82,23 @@ def get_x_y(data_frame, attr_labels, pos_label=None, survival=True, competing_ri
         Whether to return `y` that can be used for survival analysis.
 
     competing_risks : bool, optional, default: False
-        Whether `y` refers to competing risks situation. Only used if `survival` is True
+        Whether `y` refers to competing risks situation. Only used if `survival` is `True`.
 
     Returns
     -------
     X : pandas.DataFrame, shape = (n_samples, n_columns - len(attr_labels))
         Data frame containing features.
 
-    y : None or pandas.DataFrame, shape = (n_samples, len(attr_labels))
-        Data frame containing columns with supervised information.
-        If `survival` was `True`, then the column denoting the event
-        indicator will be boolean and survival times will be float.
-        If `attr_labels` contains `None`, y is set to `None`.
+    y : structured array, shape = (n_samples,), or pandas.DataFrame, shape = (n_samples, len(attr_labels)), or None
+        If `survival` is `True`, a structured array with two fields.
+        The first field is a boolean where ``True`` indicates an event and ``False``
+        indicates right-censoring. The second field is a float with the time of
+        event or time of censoring.
+
+        If `survival` is `False` and `attr_labels` not `None`, a :class:`pandas.DataFrame`
+        with columns specified by `attr_labels`.
+
+        If `survival` is `False` and `attr_labels` is `None`, `y` is set to `None`.
     """
     if survival:
         if len(attr_labels) != 2:
@@ -111,7 +116,7 @@ def _loadarff_with_index(filename):
         if isinstance(dataset["index"].dtype, CategoricalDtype):
             # concatenating categorical index may raise TypeError
             # see https://github.com/pandas-dev/pandas/issues/14586
-            dataset["index"] = dataset["index"].astype(object)
+            dataset = dataset.astype({"index": "str"})
         dataset.set_index("index", inplace=True)
     return dataset
 
@@ -154,7 +159,7 @@ def load_arff_files_standardized(
         Whether to standardize data to zero mean and unit variance.
         See :func:`sksurv.column.standardize`.
 
-    to_numeric : boo, optional, default: True
+    to_numeric : bool, optional, default: True
         Whether to convert categorical variables to numeric values.
         See :func:`sksurv.column.categorical_to_numeric`.
 
@@ -163,14 +168,34 @@ def load_arff_files_standardized(
     x_train : pandas.DataFrame, shape = (n_train, n_features)
         Training data.
 
-    y_train : pandas.DataFrame, shape = (n_train, n_labels)
+    y_train : structured array, shape = (n_train,), or pandas.DataFrame, shape = (n_train, len(attr_labels))
         Dependent variables of training data.
 
-    x_test : None or pandas.DataFrame, shape = (n_train, n_features)
+        If `survival` is `True`, a structured array with two fields.
+        The first field is a boolean where ``True`` indicates an event and ``False``
+        indicates right-censoring. The second field is a float with the time of
+        event or time of censoring.
+
+        If `survival` is `False` and `attr_labels` not `None`, a :class:`pandas.DataFrame`
+        with columns specified by `attr_labels`.
+
+        If `survival` is `False` and `attr_labels` is `None`, `y_train` is set to `None`.
+
+    x_test : None or pandas.DataFrame, shape = (n_test, n_features)
         Testing data if `path_testing` was provided.
 
-    y_test : None or pandas.DataFrame, shape = (n_train, n_labels)
+    y_test : None or structured array, shape = (n_test,)
         Dependent variables of testing data if `path_testing` was provided.
+
+        If `survival` is `True`, a structured array with two fields.
+        The first field is a boolean where ``True`` indicates an event and ``False``
+        indicates right-censoring. The second field is a float with the time of
+        event or time of censoring.
+
+        If `survival` is `False` and `attr_labels` not `None`, a :class:`pandas.DataFrame`
+        with columns specified by `attr_labels`.
+
+        If `survival` is `False` and `attr_labels` is `None`, `y_test` is set to `None`.
     """
     dataset = _loadarff_with_index(path_training)
 
@@ -237,7 +262,7 @@ def load_whas500():
 
     y : structured array with 2 fields
         *fstat*: boolean indicating whether the endpoint has been reached
-        or the event time is right censored.
+        or the event time is right-censored.
 
         *lenfol*: total length of follow-up (days from hospital admission date
         to date of last follow-up)
@@ -269,7 +294,7 @@ def load_gbsg2():
 
     y : structured array with 2 fields
         *cens*: boolean indicating whether the endpoint has been reached
-        or the event time is right censored.
+        or the event time is right-censored.
 
         *time*: total length of follow-up
 
@@ -302,7 +327,7 @@ def load_veterans_lung_cancer():
 
     y : structured array with 2 fields
         *Status*: boolean indicating whether the endpoint has been reached
-        or the event time is right censored.
+        or the event time is right-censored.
 
         *Survival_in_days*: total length of follow-up
 
@@ -328,8 +353,8 @@ def load_aids(endpoint="aids"):
 
     Parameters
     ----------
-    endpoint : aids|death
-        The endpoint
+    endpoint : {'aids', 'death'}, default: 'aids'
+        The endpoint.
 
     Returns
     -------
@@ -338,7 +363,7 @@ def load_aids(endpoint="aids"):
 
     y : structured array with 2 fields
         *censor*: boolean indicating whether the endpoint has been reached
-        or the event time is right censored.
+        or the event time is right-censored.
 
         *time*: total length of follow-up
 
@@ -384,7 +409,7 @@ def load_breast_cancer():
 
     y : structured array with 2 fields
         *e.tdm*: boolean indicating whether the endpoint has been reached
-        or the event time is right censored.
+        or the event time is right-censored.
 
         *t.tdm*: time to distant metastasis (days)
 
@@ -428,7 +453,7 @@ def load_flchain():
 
     y : structured array with 2 fields
         *death*: boolean indicating whether the subject died
-        or the event time is right censored.
+        or the event time is right-censored.
 
         *futime*: total length of follow-up or time of death.
 
@@ -473,7 +498,7 @@ def load_bmt():
         The measurements for each patient.
 
     y : structured array with 2 fields
-        *status*: Integer indicating the endpoint: 0-(survival i.e. right censored data), 1-(TRM), 2-(relapse)
+        *status*: Integer indicating the endpoint: 0-(survival i.e. right-censored data), 1-(TRM), 2-(relapse)
 
         *ftime*: total length of follow-up or time of event.
 
@@ -487,7 +512,7 @@ def load_bmt():
     """
     full_path = _get_data_path("bmt.arff")
     data = loadarff(full_path)
-    data["ftime"] = data["ftime"].astype(int)
+    data = data.astype({"ftime": int})
     return get_x_y(data, attr_labels=["status", "ftime"], competing_risks=True)
 
 
@@ -566,7 +591,7 @@ def load_cgvhd():
         The measurements for each patient.
 
     y : structured array with 2 fields
-        *status*: Integer indicating the endpoint: 0: right censored data; 1: CGVHD; 2: relapse; 3: death.
+        *status*: Integer indicating the endpoint: 0: right-censored data; 1: CGVHD; 2: relapse; 3: death.
 
         *ftime*: total length of follow-up or time of event.
 
@@ -578,8 +603,8 @@ def load_cgvhd():
     """
     full_path = _get_data_path("cgvhd.arff")
     data = loadarff(full_path)
-    data["ftime"] = data[["survtime", "reltime", "cgvhtime"]].min(axis=1)
-    data["status"] = (
+    data.loc[:, "ftime"] = data[["survtime", "reltime", "cgvhtime"]].min(axis=1)
+    data.loc[:, "status"] = (
         ((data["ftime"] == data["cgvhtime"]) & (data["cgvh"] == "1")).astype(int)
         + 2 * ((data["ftime"] == data["reltime"]) & (data["rcens"] == "1")).astype(int)
         + 3 * ((data["ftime"] == data["survtime"]) & (data["stat"] == "1")).astype(int)

sksurv/docstrings.py

@@ -0,0 +1,99 @@
+# This program is free software: you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation, either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+_PRED_SURV_FN_EXAMPLE_DOC = """
+        .. plot::
+
+            >>> import matplotlib.pyplot as plt
+            >>> from sksurv.datasets import load_veterans_lung_cancer
+            >>> from sksurv.preprocessing import OneHotEncoder
+            >>> from sksurv.{estimator_mod} import {estimator_class}
+
+            Load the data and encode categorical features.
+
+            >>> X, y = load_veterans_lung_cancer()
+            >>> Xt = OneHotEncoder().fit_transform(X)
+
+            Fit the model.
+
+            >>> estimator = {estimator_class}().fit(Xt, y)
+
+            Estimate the survival function for the first 10 samples.
+
+            >>> surv_funcs = estimator.predict_survival_function(Xt.iloc[:10])
+
+            Plot the estimated survival functions.
+
+            >>> for fn in surv_funcs:
+            ...     plt.step(fn.x, fn(fn.x), where="post")
+            ...
+            [...]
+            >>> plt.ylim(0, 1)
+            (0.0, 1.0)
+            >>> plt.show()  # doctest: +SKIP
+"""
+
+_PRED_CUMHAZ_FN_EXAMPLE_DOC = """
+        .. plot::
+
+            >>> import matplotlib.pyplot as plt
+            >>> from sksurv.datasets import load_veterans_lung_cancer
+            >>> from sksurv.preprocessing import OneHotEncoder
+            >>> from sksurv.{estimator_mod} import {estimator_class}
+
+            Load the data and encode categorical features.
+
+            >>> X, y = load_veterans_lung_cancer()
+            >>> Xt = OneHotEncoder().fit_transform(X)
+
+            Fit the model.
+
+            >>> estimator = {estimator_class}().fit(Xt, y)
+
+            Estimate the cumulative hazard function for the first 10 samples.
+
+            >>> chf_funcs = estimator.predict_cumulative_hazard_function(Xt.iloc[:10])
+
+            Plot the estimated cumulative hazard functions.
+
+            >>> for fn in chf_funcs:
+            ...     plt.step(fn.x, fn(fn.x), where="post")
+            ...
+            [...]
+            >>> plt.show()  # doctest: +SKIP
+"""
+
+
+def append_survival_function_example(*, estimator_mod, estimator_class):
+    """Append example of using predict_survival_function to API doc"""
+
+    def func(f):
+        f.__doc__ += _PRED_SURV_FN_EXAMPLE_DOC.format(
+            estimator_mod=estimator_mod,
+            estimator_class=estimator_class,
+        )
+        return f
+
+    return func
+
+
+def append_cumulative_hazard_example(*, estimator_mod, estimator_class):
+    """Append example of using predict_cumulative_hazard_function to API doc"""
+
+    def func(f):
+        f.__doc__ += _PRED_CUMHAZ_FN_EXAMPLE_DOC.format(
+            estimator_mod=estimator_mod,
+            estimator_class=estimator_class,
+        )
+        return f
+
+    return func