aplr 10.5.0__cp38-cp38-win32.whl → 10.6.0__cp38-cp38-win32.whl

aplr/aplr.py

@@ -1,6 +1,7 @@
-from typing import List, Callable, Optional, Dict
+from typing import List, Callable, Optional, Dict, Union
 import numpy as np
 import aplr_cpp
+import itertools
 
 FloatVector = np.ndarray
 FloatMatrix = np.ndarray
@@ -12,7 +13,7 @@ class APLRRegressor:
     def __init__(
         self,
         m: int = 20000,
-        v: float = 0.1,
+        v: float = 0.5,
         random_state: int = 0,
         loss_function: str = "mse",
         link_function: str = "identity",
@@ -21,9 +22,9 @@ class APLRRegressor:
         bins: int = 300,
         max_interaction_level: int = 1,
         max_interactions: int = 100000,
-        min_observations_in_split: int = 20,
-        ineligible_boosting_steps_added: int = 10,
-        max_eligible_terms: int = 5,
+        min_observations_in_split: int = 4,
+        ineligible_boosting_steps_added: int = 15,
+        max_eligible_terms: int = 7,
         verbosity: int = 0,
         dispersion_parameter: float = 1.5,
         validation_tuning_metric: str = "default",
@@ -351,7 +352,7 @@ class APLRClassifier:
     def __init__(
         self,
         m: int = 20000,
-        v: float = 0.1,
+        v: float = 0.5,
         random_state: int = 0,
         n_jobs: int = 0,
         cv_folds: int = 5,
@@ -359,9 +360,9 @@ class APLRClassifier:
         verbosity: int = 0,
         max_interaction_level: int = 1,
         max_interactions: int = 100000,
-        min_observations_in_split: int = 20,
-        ineligible_boosting_steps_added: int = 10,
-        max_eligible_terms: int = 5,
+        min_observations_in_split: int = 4,
+        ineligible_boosting_steps_added: int = 15,
+        max_eligible_terms: int = 7,
         boosting_steps_before_interactions_are_allowed: int = 0,
         monotonic_constraints_ignore_interactions: bool = False,
         early_stopping_rounds: int = 500,
@@ -531,3 +532,62 @@ class APLRClassifier:
     # For sklearn
     def predict_proba(self, X: FloatMatrix) -> FloatMatrix:
         return self.predict_class_probabilities(X)
+
+
+class APLRTuner:
+    def __init__(
+        self,
+        parameters: Union[Dict[str, List[float]], List[Dict[str, List[float]]]] = {
+            "max_interaction_level": [0, 1],
+            "min_observations_in_split": [4, 10, 20, 100, 500, 1000],
+        },
+        is_regressor: bool = True,
+    ):
+        self.parameters = parameters
+        self.is_regressor = is_regressor
+        self.parameter_grid = self._create_parameter_grid()
+
+    def _create_parameter_grid(self) -> List[Dict[str, float]]:
+        items = sorted(self.parameters.items())
+        keys, values = zip(*items)
+        combinations = list(itertools.product(*values))
+        grid = [dict(zip(keys, combination)) for combination in combinations]
+        return grid
+
+    def fit(self, X: FloatMatrix, y: FloatVector, **kwargs):
+        self.cv_results: List[Dict[str, float]] = []
+        best_validation_result = np.inf
+        for params in self.parameter_grid:
+            if self.is_regressor:
+                model = APLRRegressor(**params)
+            else:
+                model = APLRClassifier(**params)
+            model.fit(X, y, **kwargs)
+            cv_error_for_this_model = model.get_cv_error()
+            cv_results_for_this_model = model.get_params()
+            cv_results_for_this_model["cv_error"] = cv_error_for_this_model
+            self.cv_results.append(cv_results_for_this_model)
+            if cv_error_for_this_model < best_validation_result:
+                best_validation_result = cv_error_for_this_model
+                self.best_model = model
+        self.cv_results = sorted(self.cv_results, key=lambda x: x["cv_error"])
+
+    def predict(self, X: FloatMatrix, **kwargs) -> Union[FloatVector, List[str]]:
+        return self.best_model.predict(X, **kwargs)
+
+    def predict_class_probabilities(self, X: FloatMatrix, **kwargs) -> FloatMatrix:
+        if self.is_regressor == False:
+            return self.best_model.predict_class_probabilities(X, **kwargs)
+        else:
+            raise TypeError(
+                "predict_class_probabilities is only possible when is_regressor is False"
+            )
+
+    def predict_proba(self, X: FloatMatrix, **kwargs) -> FloatMatrix:
+        return self.predict_class_probabilities(X, **kwargs)
+
+    def get_best_estimator(self) -> Union[APLRClassifier, APLRRegressor]:
+        return self.best_model
+
+    def get_cv_results(self) -> List[Dict[str, float]]:
+        return self.cv_results

{aplr-10.5.0.dist-info → aplr-10.6.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: aplr
-Version: 10.5.0
+Version: 10.6.0
 Summary: Automatic Piecewise Linear Regression
 Home-page: https://github.com/ottenbreit-data-science/aplr
 Author: Mathias von Ottenbreit
@@ -19,7 +19,7 @@ Requires-Dist: numpy >=1.11
 Automatic Piecewise Linear Regression.
 
 # About
-Build predictive and interpretable parametric regression or classification machine learning models in Python based on the Automatic Piecewise Linear Regression (APLR) methodology developed by Mathias von Ottenbreit. APLR is often able to compete with tree-based methods on predictiveness, but unlike tree-based methods APLR is interpretable. Please see the [documentation](https://github.com/ottenbreit-data-science/aplr/tree/main/documentation) for more information. Links to published article: [https://link.springer.com/article/10.1007/s00180-024-01475-4](https://link.springer.com/article/10.1007/s00180-024-01475-4) and [https://rdcu.be/dz7bF](https://rdcu.be/dz7bF). More functionality has been added to APLR since the article was published.
+Build predictive and interpretable parametric regression or classification machine learning models in Python based on the Automatic Piecewise Linear Regression (APLR) methodology developed by Mathias von Ottenbreit. APLR is often able to compete with tree-based methods on predictiveness, but unlike tree-based methods APLR is interpretable. Furthermore, APLR produces smoother predictions than tree-based methods. Please see the [documentation](https://github.com/ottenbreit-data-science/aplr/tree/main/documentation) for more information. Links to published article: [https://link.springer.com/article/10.1007/s00180-024-01475-4](https://link.springer.com/article/10.1007/s00180-024-01475-4) and [https://rdcu.be/dz7bF](https://rdcu.be/dz7bF). More functionality has been added to APLR since the article was published.
 
 # How to install
 ***pip install aplr***

aplr-10.6.0.dist-info/RECORD

@@ -0,0 +1,8 @@
+aplr_cpp.cp38-win32.pyd,sha256=5bJiuXAovtUdo5QEIStpK8qjs7gxi1VGc_wSoffEKjg,509952
+aplr/__init__.py,sha256=oDFSgVytP_qQ8ilun6oHxKr-DYEeqjEQp5FciX45lls,21
+aplr/aplr.py,sha256=Qc_n1PQ1dKqyIR9UXKi73fX6HBGpmu7fayZHJxxCTXA,26006
+aplr-10.6.0.dist-info/LICENSE,sha256=YOMo-RaL4P7edMZGD96-NskKpxyMZdP3-WiiMMmihNk,1134
+aplr-10.6.0.dist-info/METADATA,sha256=orPj166YzyjPQirasMrR-AUJn8nJrulSNHdi_YUvyDg,2166
+aplr-10.6.0.dist-info/WHEEL,sha256=3szW57pNWqFVRic6Ma4o9j5UYGJmr9Fs-0ZX6rmK_Y4,95
+aplr-10.6.0.dist-info/top_level.txt,sha256=DXVC0RIFGpzVnPeKWAZTXQdJheOEZL51Wip6Fx7zbR4,14
+aplr-10.6.0.dist-info/RECORD,,

{aplr-10.5.0.dist-info → aplr-10.6.0.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (70.2.0)
+Generator: setuptools (72.1.0)
 Root-Is-Purelib: false
 Tag: cp38-cp38-win32
 

aplr-10.5.0.dist-info/RECORD

@@ -1,8 +0,0 @@
-aplr_cpp.cp38-win32.pyd,sha256=1KDaBP-eU7OzURkVc-IE1Kavl7-qpyzXi0NlA0fxLmU,509952
-aplr/__init__.py,sha256=oDFSgVytP_qQ8ilun6oHxKr-DYEeqjEQp5FciX45lls,21
-aplr/aplr.py,sha256=SHeyJWIPM_2GkD7cP-d-kPaPyki__7_wXe6SV1cYDSQ,23483
-aplr-10.5.0.dist-info/LICENSE,sha256=YOMo-RaL4P7edMZGD96-NskKpxyMZdP3-WiiMMmihNk,1134
-aplr-10.5.0.dist-info/METADATA,sha256=oISUqQCwsBAGazVcNyUL-Q58iSz09PchKv48WqwAF6Q,2093
-aplr-10.5.0.dist-info/WHEEL,sha256=bWd3bUqn9udrzjSPEGyjFzalPqRpt8KhSyToGhtYrI8,95
-aplr-10.5.0.dist-info/top_level.txt,sha256=DXVC0RIFGpzVnPeKWAZTXQdJheOEZL51Wip6Fx7zbR4,14
-aplr-10.5.0.dist-info/RECORD,,