fix set issue for pandas 0.23

examples/copula_frailty_weibull_model.py

@@ -0,0 +1,54 @@
+# -*- coding: utf-8 -*-
+from lifelines.fitters import ParametricRegressionFitter
+from autograd.scipy.special import expit
+from autograd import numpy as np
+from lifelines.utils.safe_exp import safe_exp
+
+exp = safe_exp
+dot = np.dot
+
+
+class CopulaFrailtyWeilbullModel(ParametricRegressionFitter):
+    """
+
+    Assume that subjects have two competing risks that are _not_ independent.
+    Assume that their is a frailty term that both risks depend on, and, conditional
+    on this term, the risks are IID Weibull models. If the frailty term is
+    stable with parameter alpha, then the cumulative hazard can be written down explicitly.
+
+
+
+    Reference
+    --------------
+    Frees and Valdez, UNDERSTANDING RELATIONSHIPS USING COPULAS
+
+    """
+
+    _fitted_parameter_names = ["lambda1", "rho1", "lambda2", "rho2", "alpha"]
+
+    def _cumulative_hazard(self, params, T, Xs):
+        lambda1 = exp(dot(Xs["lambda1"], params["lambda1"]))
+        lambda2 = exp(dot(Xs["lambda2"], params["lambda2"]))
+        rho2 = exp(dot(Xs["rho2"], params["rho2"]))
+        rho1 = exp(dot(Xs["rho1"], params["rho1"]))
+        alpha = exp(dot(Xs["alpha"], params["alpha"]))
+
+        return ((T / lambda1) ** rho1 + (T / lambda2) ** rho2) ** alpha
+
+
+swf = CopulaFrailtyWeilbullModel(penalizer=0.1)
+
+rossi = load_rossi()
+rossi["intercept"] = 1.0
+rossi["week"] = rossi["week"] / 54.0
+
+covariates = {
+    "lambda1": rossi.columns,
+    "lambda2": rossi.columns,
+    "rho1": ["intercept"],
+    "rho2": ["intercept"],
+    "alpha": ["intercept"],
+}
+
+swf.fit(rossi, "week", event_col="arrest", regressors=covariates, timeline=np.linspace(0, 2))
+swf.print_summary(2)

lifelines/fitters/__init__.py

@@ -1329,7 +1329,7 @@ def _filter_dataframe_to_covariates(self, df):
         cols = set(sum(self.regressors.values(), []))
         if "_intercept" not in df.columns:
             cols = cols - set(["_intercept"])
-        return df[cols]
+        return df[list(cols)]
 
     def _fit(
         self,