ngboost 0.5.2.dev0__tar.gz → 0.5.4__tar.gz

{ngboost-0.5.2.dev0 → ngboost-0.5.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ngboost
-Version: 0.5.2.dev0
+Version: 0.5.4
 Summary: Library for probabilistic predictions via gradient boosting.
 Home-page: https://github.com/stanfordmlgroup/ngboost
 License: Apache-2.0

{ngboost-0.5.2.dev0 → ngboost-0.5.4}/ngboost/ngboost.py

@@ -1,4 +1,5 @@
 """The NGBoost library"""
+
 # pylint: disable=line-too-long,too-many-instance-attributes,too-many-arguments
 # pylint: disable=unused-argument,too-many-locals,too-many-branches,too-many-statements
 # pylint: disable=unused-variable,invalid-unary-operand-type,attribute-defined-outside-init
@@ -320,29 +321,69 @@ class NGBoost:
         # This will overwrite any X_val and Y_val values passed by the user directly.
         if self.early_stopping_rounds is not None:
             early_stopping_rounds = self.early_stopping_rounds
-
-            if sample_weight is None:
-                X, X_val, Y, Y_val = train_test_split(
-                    X,
-                    Y,
-                    test_size=self.validation_fraction,
-                    random_state=self.random_state,
+            if X_val is None and Y_val is None:
+                print(
+                    f"early_stopping_rounds is set to {early_stopping_rounds} but no validation set is provided creating val set with {self.validation_fraction} of the training data"
                 )
+                if sample_weight is None:
+                    print(
+                        "Creating validation set without sample weight similar to the training data"
+                    )
+                    X, X_val, Y, Y_val = train_test_split(
+                        X,
+                        Y,
+                        test_size=self.validation_fraction,
+                        random_state=self.random_state,
+                    )
+
+                else:
+                    print(
+                        "Creating validation set with sample weight similar to the training data"
+                    )
+                    (
+                        X,
+                        X_val,
+                        Y,
+                        Y_val,
+                        sample_weight,
+                        val_sample_weight,
+                    ) = train_test_split(
+                        X,
+                        Y,
+                        sample_weight,
+                        test_size=self.validation_fraction,
+                        random_state=self.random_state,
+                    )
+            elif X_val is not None and Y_val is not None:
+                if sample_weight is not None and val_sample_weight is None:
+                    raise ValueError(
+                        "Training data is passed with sample weights but the validation data is missing sample weights pass the appropriate val_sample_weights"
+                    )
+                if sample_weight is None and val_sample_weight is not None:
+                    raise ValueError(
+                        "sample weights mismatch between training and validation data check and pass the appropriate val_sample_weights"
+                    )
+
+                print("Using passed validation data to check for early stopping.")
 
             else:
-                X, X_val, Y, Y_val, sample_weight, val_sample_weight = train_test_split(
-                    X,
-                    Y,
-                    sample_weight,
-                    test_size=self.validation_fraction,
-                    random_state=self.random_state,
-                )
+                if (X_val is not None and Y_val is None) or (
+                    X_val is None and Y_val is not None
+                ):
+                    raise ValueError(
+                        "Inconsistent Validation data either X_val or Y_val is missing, check the data"
+                    )
 
         if Y is None:
             raise ValueError("y cannot be None")
 
         X, Y = check_X_y(
-            X, Y, accept_sparse=True, y_numeric=True, multi_output=self.multi_output
+            X,
+            Y,
+            accept_sparse=True,
+            force_all_finite="allow-nan",
+            multi_output=self.multi_output,
+            y_numeric=True,
         )
 
         self.n_features = X.shape[1]
@@ -357,8 +398,9 @@ class NGBoost:
                 X_val,
                 Y_val,
                 accept_sparse=True,
-                y_numeric=True,
+                force_all_finite="allow-nan",
                 multi_output=self.multi_output,
+                y_numeric=True,
             )
             val_params = self.pred_param(X_val)
             val_loss_list = []
@@ -490,7 +532,7 @@ class NGBoost:
             A NGBoost distribution object
         """
 
-        X = check_array(X, accept_sparse=True)
+        X = check_array(X, accept_sparse=True, force_all_finite="allow-nan")
 
         params = np.asarray(self.pred_param(X, max_iter))
         dist = self.Dist(params.T)
@@ -537,7 +579,7 @@ class NGBoost:
             Numpy array of the estimates of Y
         """
 
-        X = check_array(X, accept_sparse=True)
+        X = check_array(X, accept_sparse=True, force_all_finite="allow-nan")
 
         return self.pred_dist(X, max_iter=max_iter).predict()
 

{ngboost-0.5.2.dev0 → ngboost-0.5.4}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "ngboost"
-version = "0.5.2dev"
+version = "0.5.4"
 description = "Library for probabilistic predictions via gradient boosting."
 authors = ["Stanford ML Group <avati@cs.stanford.edu>"]
 readme = "README.md"
@@ -28,6 +28,7 @@ isort = "^5.6.4"
 pylint = "^3.0.3"
 flake8 = "^7.0.0"
 
+
 [build-system]
 requires = ["poetry-core>=1.0.0"]
 build-backend = "poetry.core.masonry.api"

{ngboost-0.5.2.dev0 → ngboost-0.5.4}/setup.py

@@ -16,7 +16,7 @@ install_requires = \
 
 setup_kwargs = {
     'name': 'ngboost',
-    'version': '0.5.2.dev0',
+    'version': '0.5.4',
     'description': 'Library for probabilistic predictions via gradient boosting.',
     'long_description': '# NGBoost: Natural Gradient Boosting for Probabilistic Prediction\n\n<h4 align="center">\n\n![Python package](https://github.com/stanfordmlgroup/ngboost/workflows/Python%20package/badge.svg)\n[![GitHub Repo Size](https://img.shields.io/github/repo-size/stanfordmlgroup/ngboost?label=Repo+Size)](https://github.com/stanfordmlgroup/ngboost/graphs/contributors)\n[![Github License](https://img.shields.io/badge/License-Apache%202.0-blue.svg)](https://opensource.org/licenses/Apache-2.0)\n[![Code style: black](https://img.shields.io/badge/code%20style-black-000000.svg)](https://github.com/psf/black)\n[![PyPI](https://img.shields.io/pypi/v/ngboost?logo=pypi&logoColor=white)](https://pypi.org/project/ngboost)\n[![PyPI Downloads](https://img.shields.io/pypi/dm/ngboost?logo=icloud&logoColor=white)](https://pypistats.org/packages/ngboost)\n\n</h4>\n\nngboost is a Python library that implements Natural Gradient Boosting, as described in ["NGBoost: Natural Gradient Boosting for Probabilistic Prediction"](https://stanfordmlgroup.github.io/projects/ngboost/). It is built on top of [Scikit-Learn](https://scikit-learn.org/stable/), and is designed to be scalable and modular with respect to choice of proper scoring rule, distribution, and base learner. A didactic introduction to the methodology underlying NGBoost is available in this [slide deck](https://docs.google.com/presentation/d/1Tn23Su0ygR6z11jy3xVNiLGv0ggiUQue/edit?usp=share_link&ouid=102290675300480810195&rtpof=true&sd=true).\n\n## Installation\n\n```sh\nvia pip\n\npip install --upgrade ngboost\n\nvia conda-forge\n\nconda install -c conda-forge ngboost\n```\n\n## Usage\n\nProbabilistic regression example on the Boston housing dataset:\n\n```python\nfrom ngboost import NGBRegressor\n\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.metrics import mean_squared_error\n\n#Load Boston housing dataset\ndata_url = "http://lib.stat.cmu.edu/datasets/boston"\nraw_df = pd.read_csv(data_url, sep="\\s+", skiprows=22, header=None)\nX = np.hstack([raw_df.values[::2, :], raw_df.values[1::2, :2]])\nY = raw_df.values[1::2, 2]\n\nX_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size=0.2)\n\nngb = NGBRegressor().fit(X_train, Y_train)\nY_preds = ngb.predict(X_test)\nY_dists = ngb.pred_dist(X_test)\n\n# test Mean Squared Error\ntest_MSE = mean_squared_error(Y_preds, Y_test)\nprint(\'Test MSE\', test_MSE)\n\n# test Negative Log Likelihood\ntest_NLL = -Y_dists.logpdf(Y_test).mean()\nprint(\'Test NLL\', test_NLL)\n```\n\nDetails on available distributions, scoring rules, learners, tuning, and model interpretation are available in our [user guide](https://stanfordmlgroup.github.io/ngboost/intro.html), which also includes numerous usage examples and information on how to add new distributions or scores to NGBoost.\n\n## License\n\n[Apache License 2.0](https://github.com/stanfordmlgroup/ngboost/blob/master/LICENSE).\n\n## Reference\n\nTony Duan, Anand Avati, Daisy Yi Ding, Khanh K. Thai, Sanjay Basu, Andrew Y. Ng, Alejandro Schuler. 2019.\nNGBoost: Natural Gradient Boosting for Probabilistic Prediction.\n[arXiv](https://arxiv.org/abs/1910.03225)\n',
     'author': 'Stanford ML Group',