Merge pull request #8 from jzluo/endometrial

Endometrial dataset

.flake8

@@ -1,3 +1,4 @@
 [flake8]
 max-line-length = 88
 extend-ignore = E203
+per-file-ignores = firthlogist/__init__.py:F401

firthlogist/__init__.py

@@ -1 +1,5 @@
-from .firthlogist import FirthLogisticRegression, load_sex2  # noqa F401
+from .firthlogist import (  # noqa F401
+    FirthLogisticRegression,
+    load_endometrial,
+    load_sex2,
+)

firthlogist/datasets/endometrial.csv

@@ -0,0 +1,80 @@
+"NV","PI","EH","HG"
+0,13,1.64,0
+0,16,2.26,0
+0,8,3.14,0
+0,34,2.68,0
+0,20,1.28,0
+0,5,2.31,0
+0,17,1.8,0
+0,10,1.68,0
+0,26,1.56,0
+0,17,2.31,0
+0,8,2.01,0
+0,7,1.89,0
+0,20,3.15,0
+0,10,1.23,0
+0,18,1.27,0
+0,16,1.76,0
+0,18,2,0
+0,8,2.64,1
+0,29,0.88,1
+0,12,1.27,1
+0,20,1.37,1
+1,38,0.97,1
+1,22,1.14,1
+1,7,0.88,1
+1,25,0.91,1
+1,15,0.58,1
+0,7,0.97,1
+0,28,1.5,0
+0,11,1.33,0
+0,19,2.37,0
+0,10,1.82,0
+0,10,3.13,0
+0,18,1.31,0
+0,14,1.92,0
+0,21,1.64,0
+0,11,2.01,0
+0,17,1.88,0
+0,25,1.93,0
+0,16,2.11,0
+0,19,1.29,0
+0,15,1.72,0
+0,33,0.75,0
+0,24,1.92,0
+0,48,1.84,1
+0,12,1.11,1
+0,19,1.61,1
+0,2,1.18,1
+1,22,1.44,1
+1,40,1.18,1
+1,5,0.93,1
+1,0,1.17,1
+0,21,1.19,1
+0,15,1.06,1
+0,29,2.02,0
+0,15,2.29,0
+0,12,2.33,0
+0,3,2.9,0
+0,20,1.7,0
+0,23,1.41,0
+0,12,2.25,0
+0,22,1.54,0
+0,42,1.97,0
+0,15,1.75,0
+0,13,2.16,0
+0,14,2.57,0
+0,19,1.37,0
+0,12,3.61,0
+0,13,2.04,0
+0,10,2.17,0
+0,12,1.69,1
+1,49,0.27,1
+0,6,1.84,1
+0,5,1.3,1
+0,17,0.96,1
+1,11,1.01,1
+1,21,0.98,1
+0,5,0.35,1
+1,19,1.02,1
+0,33,0.85,1

firthlogist/firthlogist.py

@@ -83,10 +83,10 @@ class FirthLogisticRegression(BaseEstimator, ClassifierMixin):
     def __init__(
         self,
         max_iter=25,
-        max_halfstep=25,
+        max_halfstep=0,
         max_stepsize=5,
         pl_max_iter=100,
-        pl_max_halfstep=25,
+        pl_max_halfstep=0,
         pl_max_stepsize=5,
         tol=0.0001,
         fit_intercept=True,
@@ -444,3 +444,34 @@ def load_sex2():
     X = X[:, 1:]
     feature_names = ["age", "oc", "vic", "vicl", "vis", "dia"]
     return X, y, feature_names
+
+
+def load_endometrial():
+    """
+    Load the endometrial cancer dataset analyzed in Heinze and Schemper (2002). The data
+    was originally provided by Dr E. Asseryanis from the Vienna University Medical
+    School
+
+    Returns
+    -------
+    X
+        endometrial data as numpy array
+    y
+        endometrial `HG` target column
+    feature_names
+        List of feature names
+
+    References
+    ----------
+    Agresti, A (2015). Foundations of Linear and Generalized Linear Models.
+    Wiley Series in Probability and Statistics.
+
+    Heinze G, Schemper M (2002). A solution to the problem of separation in logistic
+    regression. Statistics in Medicine 21: 2409-2419.
+    """
+    with open_text("firthlogist.datasets", "endometrial.csv") as sex2:
+        X = np.loadtxt(sex2, skiprows=1, delimiter=",")
+    y = X[:, -1]
+    X = X[:, :-1]
+    feature_names = ["NV", "PI", "EH"]
+    return X, y, feature_names

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "firthlogist"
-version = "0.3.0"
+version = "0.3.1"
 description = "Python implementation of Logistic Regression with Firth's bias reduction"
 authors = ["Jon Luo <[email protected]>"]
 repository = "https://github.com/jzluo/firthlogist"