Merge pull request #231 from guillermo-navas-palencia/develop

Develop
guillermo-navas-palencia · Feb 12, 2023 · 9735aca · 9735aca
2 parents d20ed46 + da700e2
commit 9735aca
Show file tree

Hide file tree

Showing 11 changed files with 112 additions and 40 deletions.
diff --git a/README.rst b/README.rst
@@ -74,7 +74,7 @@ OptBinning requires
 * numpy (>=1.16.1)
 * ortools (>=9.4)
 * pandas
-* ropwr (>=0.4.0)
+* ropwr (>=1.0.0)
 * scikit-learn (>=1.0.2)
 * scipy (>=1.6.0)
 
@@ -227,8 +227,8 @@ Print overview information about the options settings, problem statistics, and t
 
 .. code-block:: text
 
-   optbinning (Version 0.17.0)
-   Copyright (c) 2019-2022 Guillermo Navas-Palencia, Apache License 2.0
+   optbinning (Version 0.17.3)
+   Copyright (c) 2019-2023 Guillermo Navas-Palencia, Apache License 2.0
 
      Begin options
        name                         mean radius   * U
@@ -395,8 +395,8 @@ and the number of selected variables after the binning process.
 
 .. code-block:: text
 
-   optbinning (Version 0.17.0)
-   Copyright (c) 2019-2022 Guillermo Navas-Palencia, Apache License 2.0
+   optbinning (Version 0.17.3)
+   Copyright (c) 2019-2023 Guillermo Navas-Palencia, Apache License 2.0
 
      Begin options
        binning_process                      yes   * U

diff --git a/doc/source/conf.py b/doc/source/conf.py
@@ -18,13 +18,13 @@
 # -- Project information -----------------------------------------------------
 
 project = 'optbinning'
-copyright = '2019 - 2022, Guillermo Navas-Palencia'
+copyright = '2019 - 2023, Guillermo Navas-Palencia'
 author = 'Guillermo Navas-Palencia'
 
 # The short X.Y version
-version = '0.17.2'
+version = '0.17.3'
 # The full version, including alpha/beta/rc tags
-release = '0.17.2'
+release = '0.17.3'
 
 
 # -- General configuration ---------------------------------------------------

diff --git a/doc/source/release_notes.rst b/doc/source/release_notes.rst
@@ -1,6 +1,22 @@
 Release Notes
 =============
 
+Version 0.17.3 (2023-02-12)
+---------------------------
+
+Improvements:
+
+   - Implement ``sample_weight`` check in Scorecard class (`Issue 228 <https://github.com/guillermo-navas-palencia/optbinning/issues/228>`_).
+
+Bugfixes:
+
+   - Fix ``metric_missing`` ignored in Scorecard class (`Issue 226 <https://github.com/guillermo-navas-palencia/optbinning/issues/226>`_).
+
+Dependencies:
+
+   - Update RoPWR required version.
+
+
 Version 0.17.2 (2022-12-15)
 ---------------------------
 

diff --git a/optbinning/_version.py b/optbinning/_version.py
@@ -1,3 +1,3 @@
 """Version information."""
 
-__version__ = "0.17.2"
+__version__ = "0.17.3"
diff --git a/optbinning/binning/piecewise/base.py b/optbinning/binning/piecewise/base.py
@@ -28,9 +28,9 @@
 
 
 def _check_parameters(name, estimator, objective, degree, continuous,
-                      prebinning_method, max_n_prebins, min_prebin_size,
-                      min_n_bins, max_n_bins, min_bin_size, max_bin_size,
-                      monotonic_trend, n_subsamples, max_pvalue,
+                      continuous_deriv, prebinning_method, max_n_prebins,
+                      min_prebin_size, min_n_bins, max_n_bins, min_bin_size,
+                      max_bin_size, monotonic_trend, n_subsamples, max_pvalue,
                       max_pvalue_policy, outlier_detector, outlier_params,
                       user_splits, user_splits_fixed, special_codes,
                       split_digits, solver, h_epsilon, quantile,
@@ -64,6 +64,10 @@ def _check_parameters(name, estimator, objective, degree, continuous,
         raise TypeError("continuous must be a boolean; got {}."
                         .format(verbose))
 
+    if not isinstance(continuous_deriv, bool):
+        raise TypeError("continuous_deriv must be a boolean; got {}."
+                        .format(verbose))
+
     if prebinning_method not in ("cart", "quantile", "uniform"):
         raise ValueError('Invalid value for prebinning_method. Allowed string '
                          'values are "cart", "quantile" and "uniform".')
@@ -209,7 +213,8 @@ def _check_parameters(name, estimator, objective, degree, continuous,
 
 class BasePWBinning(Base, BaseEstimator):
     def __init__(self, name="", estimator=None, objective="l2", degree=1,
-                 continuous=True, prebinning_method="cart", max_n_prebins=20,
+                 continuous=True, continuous_deriv=True,
+                 prebinning_method="cart", max_n_prebins=20,
                  min_prebin_size=0.05, min_n_bins=None, max_n_bins=None,
                  min_bin_size=None, max_bin_size=None, monotonic_trend="auto",
                  n_subsamples=None, max_pvalue=None,
@@ -224,6 +229,7 @@ def __init__(self, name="", estimator=None, objective="l2", degree=1,
         self.objective = objective
         self.degree = degree
         self.continuous = continuous
+        self.continuous_deriv = continuous_deriv
         self.prebinning_method = prebinning_method
 
         self.max_n_prebins = max_n_prebins
@@ -451,9 +457,19 @@ def _fit_binning(self, x, y, prediction, lb, ub):
         time_solver = time.perf_counter()
 
         optimizer = RobustPWRegression(
-            self.objective, self.degree, self.continuous, monotonic,
-            self.solver, self.h_epsilon, self.quantile, self.regularization,
-            self.reg_l1, self.reg_l1, self.verbose)
+            objective=self.objective,
+            degree=self.degree,
+            continuous=self.continuous,
+            continuous_deriv=self.continuous_deriv,
+            monotonic_trend=monotonic,
+            solver=self.solver,
+            h_epsilon=self.h_epsilon,
+            quantile=self.quantile,
+            regularization=self.regularization,
+            reg_l1=self.reg_l1,
+            reg_l2=self.reg_l2,
+            extrapolation="continue",
+            verbose=self.verbose)
 
         optimizer.fit(x_subsamples, pred_subsamples, splits, lb=lb, ub=ub)
 

diff --git a/optbinning/binning/piecewise/binning.py b/optbinning/binning/piecewise/binning.py
@@ -55,6 +55,11 @@ class OptimalPWBinning(BasePWBinning):
     continuous : bool (default=True)
         Whether to fit a continuous or discontinuous piecewise regression.
 
+    continuous_deriv : bool (default=True)
+        Whether to fit a polynomial with continuous derivatives. This option
+        fits a smooth degree d-polynomial with d-1 continuity in derivatives
+        (splines).
+
     prebinning_method : str, optional (default="cart")
         The pre-binning method. Supported methods are "cart" for a CART
         decision tree, "quantile" to generate prebins with approximately same
@@ -175,7 +180,8 @@ class OptimalPWBinning(BasePWBinning):
         Enable verbose output.
     """
     def __init__(self, name="", estimator=None, objective="l2", degree=1,
-                 continuous=True, prebinning_method="cart", max_n_prebins=20,
+                 continuous=True, continuous_deriv=True,
+                 prebinning_method="cart", max_n_prebins=20,
                  min_prebin_size=0.05, min_n_bins=None, max_n_bins=None,
                  min_bin_size=None, max_bin_size=None, monotonic_trend="auto",
                  n_subsamples=None, max_pvalue=None,
@@ -186,13 +192,14 @@ def __init__(self, name="", estimator=None, objective="l2", degree=1,
                  reg_l2=1.0, random_state=None, verbose=False):
 
         super().__init__(name, estimator, objective, degree, continuous,
-                         prebinning_method, max_n_prebins, min_prebin_size,
-                         min_n_bins, max_n_bins, min_bin_size, max_bin_size,
-                         monotonic_trend, n_subsamples, max_pvalue,
-                         max_pvalue_policy, outlier_detector, outlier_params,
-                         user_splits, user_splits_fixed, special_codes,
-                         split_digits, solver, h_epsilon, quantile,
-                         regularization, reg_l1, reg_l2, random_state, verbose)
+                         continuous_deriv, prebinning_method, max_n_prebins,
+                         min_prebin_size, min_n_bins, max_n_bins, min_bin_size,
+                         max_bin_size, monotonic_trend, n_subsamples,
+                         max_pvalue, max_pvalue_policy, outlier_detector,
+                         outlier_params, user_splits, user_splits_fixed,
+                         special_codes, split_digits, solver, h_epsilon,
+                         quantile, regularization, reg_l1, reg_l2,
+                         random_state, verbose)
 
         self._problem_type = "classification"
 

diff --git a/optbinning/binning/piecewise/continuous_binning.py b/optbinning/binning/piecewise/continuous_binning.py
@@ -45,6 +45,11 @@ class ContinuousOptimalPWBinning(BasePWBinning):
     continuous : bool (default=True)
         Whether to fit a continuous or discontinuous piecewise regression.
 
+    continuous_deriv : bool (default=True)
+        Whether to fit a polynomial with continuous derivatives. This option
+        fits a smooth degree d-polynomial with d-1 continuity in derivatives
+        (splines).
+
     prebinning_method : str, optional (default="cart")
         The pre-binning method. Supported methods are "cart" for a CART
         decision tree, "quantile" to generate prebins with approximately same
@@ -164,25 +169,26 @@ class ContinuousOptimalPWBinning(BasePWBinning):
     verbose : bool (default=False)
         Enable verbose output.
     """
-    def __init__(self, name="", objective="l2", degree=1,
-                 continuous=True, prebinning_method="cart", max_n_prebins=20,
-                 min_prebin_size=0.05, min_n_bins=None, max_n_bins=None,
-                 min_bin_size=None, max_bin_size=None, monotonic_trend="auto",
-                 n_subsamples=None, max_pvalue=None,
+    def __init__(self, name="", objective="l2", degree=1, continuous=True,
+                 continuous_deriv=True, prebinning_method="cart",
+                 max_n_prebins=20, min_prebin_size=0.05, min_n_bins=None,
+                 max_n_bins=None, min_bin_size=None, max_bin_size=None,
+                 monotonic_trend="auto", n_subsamples=None, max_pvalue=None,
                  max_pvalue_policy="consecutive", outlier_detector=None,
                  outlier_params=None, user_splits=None, user_splits_fixed=None,
                  special_codes=None, split_digits=None, solver="auto",
                  h_epsilon=1.35, quantile=0.5, regularization=None, reg_l1=1.0,
                  reg_l2=1.0, random_state=None, verbose=False):
 
         super().__init__(name, None, objective, degree, continuous,
-                         prebinning_method, max_n_prebins, min_prebin_size,
-                         min_n_bins, max_n_bins, min_bin_size, max_bin_size,
-                         monotonic_trend, n_subsamples, max_pvalue,
-                         max_pvalue_policy, outlier_detector, outlier_params,
-                         user_splits, user_splits_fixed, special_codes,
-                         split_digits, solver, h_epsilon, quantile,
-                         regularization, reg_l1, reg_l2, random_state, verbose)
+                         continuous_deriv, prebinning_method, max_n_prebins,
+                         min_prebin_size, min_n_bins, max_n_bins, min_bin_size,
+                         max_bin_size, monotonic_trend, n_subsamples,
+                         max_pvalue, max_pvalue_policy, outlier_detector,
+                         outlier_params, user_splits, user_splits_fixed,
+                         special_codes, split_digits, solver, h_epsilon,
+                         quantile, regularization, reg_l1, reg_l2,
+                         random_state, verbose)
 
         self._problem_type = "regression"
 

diff --git a/optbinning/scorecard/scorecard.py b/optbinning/scorecard/scorecard.py
@@ -557,7 +557,11 @@ def _fit(self, X, y, sample_weight, metric_special, metric_missing,
             logger.info("Fitting estimator.")
 
         self.estimator_ = clone(self.estimator)
-        self.estimator_.fit(X_t, y, sample_weight)
+
+        if sample_weight is not None:
+            self.estimator_.fit(X_t, y, sample_weight=sample_weight)
+        else:
+            self.estimator_.fit(X_t, y)
 
         self._time_estimator = time.perf_counter() - time_estimator
 
@@ -601,7 +605,7 @@ def _fit(self, X, y, sample_weight, metric_special, metric_missing,
 
                 binning_table.loc[
                     nt-1-n_specials:nt-2, "Points"] = metric_special * c
-            elif metric_missing != 'empirical':
+            if metric_missing != 'empirical':
                 binning_table.loc[nt-1, "Points"] = metric_missing * c
 
             binning_table.index.names = ['Bin id']

diff --git a/requirements.txt b/requirements.txt
@@ -2,6 +2,6 @@ matplotlib
 numpy>=1.16.1
 ortools>=9.4
 pandas
-ropwr>=0.4.0
+ropwr>=1.0.0
 scikit-learn>=1.0.2
 scipy>=1.6.0
diff --git a/setup.py b/setup.py
@@ -54,7 +54,7 @@ def run_tests(self):
     'numpy>=1.16.1',
     'ortools>=9.4',
     'pandas',
-    'ropwr>=0.4.0',
+    'ropwr>=1.0.0',
     'scikit-learn>=1.0.2',
     'scipy>=1.6.0',
 ]

diff --git a/tests/test_scorecard.py b/tests/test_scorecard.py
@@ -450,3 +450,26 @@ def test_verbose():
     with open("tests/results/test_scorecard_verbose.txt", "w") as f:
         with redirect_stdout(f):
             scorecard.fit(X, y)
+
+
+def test_missing_metrics():
+    data = pd.DataFrame(
+        {'target': np.hstack(
+            (np.tile(np.array([0, 1]), 50),
+             np.array([0]*90 + [1]*10)
+             )
+         ),
+         'var': [np.nan] * 100 + ['A'] * 100}
+    )
+
+    binning_process = BinningProcess(['var'])
+    scaling_method_params = {'min': 0, 'max': 100}
+
+    scorecard = Scorecard(
+        binning_process=binning_process,
+        estimator=LogisticRegression(),
+        scaling_method="min_max",
+        scaling_method_params=scaling_method_params
+    ).fit(data, data.target)
+
+    assert scorecard.table()['Points'].iloc[-1] == approx(0, rel=1e-6)