Added Examples Folder and MSE Example

automatminer/automl/adaptors.py

@@ -118,9 +118,9 @@ def fit(self, df, target, **fit_kwargs):
         self._features = df.drop(columns=target).columns.tolist()
         self._ml_data = {"X": X, "y": y}
         self.fitted_target = target
-        self.logger.info("TPOT fitting started.")
+        self._logger.info("TPOT fitting started.")
         self._backend = self._backend.fit(X, y, **fit_kwargs)
-        self.logger.info("TPOT fitting finished.")
+        self._logger.info("TPOT fitting finished.")
         return self
 
     @property
@@ -219,7 +219,7 @@ def predict(self, df, target):
             X = df[self._features].values  # rectify feature order
             y_pred = self._backend.predict(X)
             df[target + " predicted"] = y_pred
-            self.logger.debug("Prediction finished successfully.")
+            self._logger.debug("Prediction finished successfully.")
             return df
 
 

automatminer/examples/mse_example.py

@@ -0,0 +1,45 @@
+import unittest
+import os
+
+from automatminer.pipeline import MatPipe
+from automatminer.presets import get_preset_config
+from matminer.datasets.dataset_retrieval import load_dataset
+from sklearn.metrics.regression import mean_squared_error
+
+
+@unittest.skipIf("CI" in os.environ.keys(), "Test too intensive for CircleCI.")
+class MSE_Example(unittest.TestCase):
+
+    """
+    The following example uses the elastic_tensor_2015 dataset and a
+    default config to create a MatPipe. This MatPipe is used to benchmark
+    the target property K_VRH.
+
+    The unit tests confirm that the output of the benchmark is not empty.
+    They also ensure that, based on this specific example, the mean
+    squared error is between 0 and 500.
+
+    For debugging purposes, you can use the debug config instead. In
+    addition, make the range of the mean squared error be 0 - 1000 rather
+    than 0 - 500.
+
+    """
+
+    def test_mse_example(self):
+        df = load_dataset("elastic_tensor_2015")
+        default_config = get_preset_config("default")
+        pipe = MatPipe(**default_config)
+        df = df.rename(columns={"formula": "composition"})[["composition", "structure", "K_VRH"]]
+        predicted = pipe.benchmark(df, "K_VRH", test_spec=0.2)
+        self.assertTrue(not predicted.empty)
+
+        y_true = predicted["K_VRH"]
+        y_test = predicted["K_VRH predicted"]
+        mse = mean_squared_error(y_true, y_test)
+        print("MSE: " + str(mse))
+        self.assertTrue(mse < 500)
+        self.assertTrue(mse > 0)
+
+
+if __name__ == '__main__':
+    unittest.main()