Merge pull request #14 from Project-Resilience/remove-cao

Removed references to CAO

src/prsdk/persistence/serializers/neural_network_serializer.py

@@ -7,7 +7,6 @@
 import joblib
 import torch
 
-from data.cao_mapping import CAOMapping
 from persistence.serializers.serializer import Serializer
 from predictors.neural_network.torch_neural_net import TorchNeuralNet
 from predictors.neural_network.neural_net_predictor import NeuralNetPredictor
@@ -31,9 +30,6 @@ def save(self, model: NeuralNetPredictor, path: Path):
 
         # Note: we don't save the model's device, as it's not guaranteed to be available on load
         config = {
-            "context": model.cao.context,
-            "actions": model.cao.actions,
-            "outcomes": model.cao.outcomes,
             "features": model.features,
             "label": model.label,
             "hidden_sizes": model.hidden_sizes,
@@ -68,9 +64,7 @@ def load(self, path: Path) -> NeuralNetPredictor:
         # Initialize model with config
         with open(path / "config.json", "r", encoding="utf-8") as file:
             config = json.load(file)
-        # Grab CAO out of config
-        cao = CAOMapping(config.pop("context"), config.pop("actions"), config.pop("outcomes"))
-        nnp = NeuralNetPredictor(cao, config)
+        nnp = NeuralNetPredictor(config)
 
         nnp.model = TorchNeuralNet(len(config["features"]),
                                    config["hidden_sizes"],

src/prsdk/persistence/serializers/sklearn_serializer.py

@@ -6,7 +6,6 @@
 
 import joblib
 
-from data.cao_mapping import CAOMapping
 from persistence.serializers.serializer import Serializer
 from predictors.sklearn_predictors.sklearn_predictor import SKLearnPredictor
 
@@ -24,13 +23,8 @@ def save(self, model: SKLearnPredictor, path: Path):
         """
         path.mkdir(parents=True, exist_ok=True)
 
-        # Add CAO to the config
-        config = dict(model.config.items())
-        cao_dict = {"context": model.cao.context, "actions": model.cao.actions, "outcomes": model.cao.outcomes}
-        config.update(cao_dict)
-
         with open(path / "config.json", "w", encoding="utf-8") as file:
-            json.dump(config, file)
+            json.dump(model.config, file)
         joblib.dump(model.model, path / "model.joblib")
 
     def load(self, path: Path) -> "SKLearnPredictor":
@@ -44,11 +38,9 @@ def load(self, path: Path) -> "SKLearnPredictor":
         if not (load_path / "config.json").exists() or not (load_path / "model.joblib").exists():
             raise FileNotFoundError("Model files not found in path.")
 
-        # Extract CAO from config
         with open(load_path / "config.json", "r", encoding="utf-8") as file:
             config = json.load(file)
-        cao = CAOMapping(config.pop("context"), config.pop("actions"), config.pop("outcomes"))
 
         model = joblib.load(load_path / "model.joblib")
-        sklearn_predictor = SKLearnPredictor(cao, model, config)
+        sklearn_predictor = SKLearnPredictor(model, config)
         return sklearn_predictor

src/prsdk/predictors/neural_network/neural_net_predictor.py

@@ -14,7 +14,6 @@
 from torch.utils.data import DataLoader
 from torch.utils.tensorboard import SummaryWriter
 
-from data.cao_mapping import CAOMapping
 from data.torch_data import TorchDataset
 from predictors.predictor import Predictor
 from predictors.neural_network.torch_neural_net import TorchNeuralNet
@@ -29,11 +28,8 @@ class NeuralNetPredictor(Predictor):
     Data is automatically standardized and the scaler is saved with the model.
     TODO: We want to be able to have custom scaling in the future.
     """
-    def __init__(self, cao: CAOMapping, model_config: dict):
+    def __init__(self, model_config: dict):
         """
-        :param context: list of context features.
-        :param actions: list of action features.
-        :param outcomes: list of outcomes to predict.
         :param model_config: dictionary of model configuration parameters.
             Model config should contain the following:
             features: list of features to use in the model (optional, defaults to all context + actions)
@@ -48,7 +44,7 @@ def __init__(self, cao: CAOMapping, model_config: dict):
             train_pct: percentage of training data to use (defaults to 1)
             step_lr_params: dictionary of parameters to pass to the step learning rate scheduler (defaults to 1, 0.1)
         """
-        super().__init__(cao)
+        super().__init__()
         self.features = model_config.get("features", None)
         self.label = model_config.get("label", None)
 

src/prsdk/predictors/predictor.py

@@ -1,26 +1,17 @@
 """
-Abstract class for predictors to inherit from.
+Interface for predictors to implement.
 """
 from abc import ABC, abstractmethod
 
 import pandas as pd
 
-from data.cao_mapping import CAOMapping
-
 
 class Predictor(ABC):
     """
-    Abstract class for predictors to inherit from.
+    Interface for predictors to implement.
     Predictors must be able to be fit and predict on a DataFrame.
     It is up to the Predictor to keep track of the proper label to label the output DataFrame.
     """
-    def __init__(self, cao: CAOMapping):
-        """
-        Initializes the Predictor with the context, actions, and outcomes.
-        :param cao: CAOMapping object with context, actions, and outcomes.
-        """
-        self.cao = cao
-
     @abstractmethod
     def fit(self, X_train: pd.DataFrame, y_train: pd.Series):
         """

src/prsdk/predictors/sklearn_predictors/linear_regression_predictor.py

@@ -3,7 +3,6 @@
 """
 from sklearn.linear_model import LinearRegression
 
-from data.cao_mapping import CAOMapping
 from predictors.sklearn_predictors.sklearn_predictor import SKLearnPredictor
 
 
@@ -12,14 +11,13 @@ class LinearRegressionPredictor(SKLearnPredictor):
     Simple linear regression predictor.
     See SKLearnPredictor for more details.
     """
-    def __init__(self, cao: CAOMapping, model_config: dict):
+    def __init__(self, model_config: dict):
         """
-        :param cao: CAOMapping object with context, actions, and outcomes for super constructor.
         :param model_config: Configuration to pass into the SKLearn constructor. Also contains the keys "features" and
             "label" to keep track of the features and label to predict.
         """
         if not model_config:
             model_config = {}
         lr_config = {key: value for key, value in model_config.items() if key not in ["features", "label"]}
         model = LinearRegression(**lr_config)
-        super().__init__(cao, model, model_config)
+        super().__init__(model, model_config)

src/prsdk/predictors/sklearn_predictors/random_forest_predictor.py

@@ -3,7 +3,6 @@
 """
 from sklearn.ensemble import RandomForestRegressor
 
-from data.cao_mapping import CAOMapping
 from predictors.sklearn_predictors.sklearn_predictor import SKLearnPredictor
 
 
@@ -12,12 +11,11 @@ class RandomForestPredictor(SKLearnPredictor):
     Simple random forest predictor.
     See SKLearnPredictor for more details.
     """
-    def __init__(self, cao: CAOMapping, model_config: dict):
+    def __init__(self, model_config: dict):
         """
-        :param cao: CAOMapping object with context, actions, and outcomes for super constructor.
         :param model_config: Configuration to pass into the SKLearn constructor. Also contains the keys "features" and
             "label" to keep track of the features and label to predict.
         """
         rf_config = {key: value for key, value in model_config.items() if key not in ["features", "label"]}
         model = RandomForestRegressor(**rf_config)
-        super().__init__(cao, model, model_config)
+        super().__init__(model, model_config)

src/prsdk/predictors/sklearn_predictors/sklearn_predictor.py

@@ -6,7 +6,6 @@
 
 import pandas as pd
 
-from data.cao_mapping import CAOMapping
 from predictors.predictor import Predictor
 
 
@@ -15,14 +14,14 @@ class SKLearnPredictor(Predictor, ABC):
     Simple abstract class for sklearn predictors.
     Keeps track of features fit on and label to predict.
     """
-    def __init__(self, cao: CAOMapping, model, model_config: dict):
+    def __init__(self, model, model_config: dict):
         """
         Model config contains the following:
         features: list of features to use for prediction (optional, defaults to all features)
         label: name of the label to predict (optional, defaults to passed label during fit)
         Any other parameters are passed to the model.
         """
-        super().__init__(cao)
+        super().__init__()
         self.config = model_config
         self.model = model
 

src/prsdk/prescriptors/prescriptor.py

@@ -1,26 +1,16 @@
 """
-Abstract prescriptor class to be implemented.
+Interface prescriptor to be implemented.
 """
 from abc import ABC, abstractmethod
 
 import pandas as pd
 
-from data.cao_mapping import CAOMapping
-
 
 # pylint: disable=too-few-public-methods
 class Prescriptor(ABC):
     """
-    Abstract class for prescriptors to allow us to experiment with different implementations.
+    Interface for prescriptors to implement.
     """
-    def __init__(self, cao: CAOMapping):
-        """
-        We keep track of the context, actions, and outcomes in the CAO mapping to ensure the prescriptor is compatible
-        with the project it's in.
-        :param cao: CAOMapping object with context, actions, and outcomes.
-        """
-        self.cao = cao
-
     @abstractmethod
     def prescribe(self, context_df: pd.DataFrame) -> pd.DataFrame:
         """

tests/persistence/test_hf_persistence.py

@@ -17,7 +17,6 @@ class TestHuggingFacePersistence(unittest.TestCase):
     """
     Tests the HuggingFace Persistor. We can't test the actual upload but we can test the download with an
     arbitrary model from HuggingFace.
-    TODO: We have to update our models to match the new configs that save CAO
     """
     def setUp(self):
         self.temp_dir = Path("tests/temp")

tests/persistence/test_predictor_serialization.py

@@ -7,7 +7,6 @@
 
 import pandas as pd
 
-from data.cao_mapping import CAOMapping
 from persistence.serializers.neural_network_serializer import NeuralNetSerializer
 from persistence.serializers.sklearn_serializer import SKLearnSerializer
 from predictors.neural_network.neural_net_predictor import NeuralNetPredictor
@@ -25,7 +24,6 @@ def setUp(self):
         2 models with the same parameters, load one from the other's save, and check if
         their predictions are the same.
         """
-        self.cao = CAOMapping(["a", "b"], ["c"], ["label"])
         self.models = [
             NeuralNetPredictor,
             LinearRegressionPredictor,
@@ -56,7 +54,7 @@ def test_save_file_names(self):
         ]
         for model, serializer, config, test_names in zip(self.models, self.serializers, self.configs, save_file_names):
             with self.subTest(model=model):
-                predictor = model(self.cao, config)
+                predictor = model(config)
                 predictor.fit(self.dummy_data, self.dummy_target)
                 serializer.save(predictor, self.temp_path)
                 files = [f.name for f in self.temp_path.glob("**/*") if f.is_file()]
@@ -71,7 +69,7 @@ def test_loaded_same(self):
         """
         for model, serializer, config in zip(self.models, self.serializers, self.configs):
             with self.subTest(model=model):
-                predictor = model(self.cao, config)
+                predictor = model(config)
                 predictor.fit(self.dummy_data.iloc[:2], self.dummy_target.iloc[:2])
                 output = predictor.predict(self.dummy_data.iloc[2:])
                 serializer.save(predictor, self.temp_path)

tests/predictors/test_neural_net.py

@@ -5,22 +5,18 @@
 
 import pandas as pd
 
-from data.cao_mapping import CAOMapping
 from predictors.neural_network.neural_net_predictor import NeuralNetPredictor
 
 
 class TestNeuralNet(unittest.TestCase):
     """
     Specifically tests the neural net predictor
     """
-    def setUp(self):
-        self.cao = CAOMapping(["a", "b"], ["c"], ["label"])
-
     def test_single_input(self):
         """
         Tests the neural net with a single input.
         """
-        predictor = NeuralNetPredictor(self.cao, {"hidden_sizes": [4], "epochs": 1, "batch_size": 1, "device": "cpu"})
+        predictor = NeuralNetPredictor({"hidden_sizes": [4], "epochs": 1, "batch_size": 1, "device": "cpu"})
 
         train_data = pd.DataFrame({"a": [1], "b": [2], "c": [3], "label": [4]})
         test_data = pd.DataFrame({"a": [4], "b": [5], "c": [6]})
@@ -33,7 +29,7 @@ def test_multi_input(self):
         """
         Tests the neural net with multiple inputs.
         """
-        predictor = NeuralNetPredictor(self.cao, {"hidden_sizes": [4], "epochs": 1, "batch_size": 1, "device": "cpu"})
+        predictor = NeuralNetPredictor({"hidden_sizes": [4], "epochs": 1, "batch_size": 1, "device": "cpu"})
 
         train_data = pd.DataFrame({"a": [1, 2], "b": [2, 3], "c": [3, 4], "label": [4, 5]})
         test_data = pd.DataFrame({"a": [4, 5], "b": [5, 6], "c": [6, 7]})
@@ -46,7 +42,7 @@ def test_batched_input(self):
         """
         Tests the neural network with batched inputs.
         """
-        predictor = NeuralNetPredictor(self.cao, {"hidden_sizes": [4], "epochs": 1, "batch_size": 2, "device": "cpu"})
+        predictor = NeuralNetPredictor({"hidden_sizes": [4], "epochs": 1, "batch_size": 2, "device": "cpu"})
 
         train_data = pd.DataFrame({"a": [1, 2, 3], "b": [2, 3, 4], "c": [3, 4, 5], "label": [4, 5, 6]})
         test_data = pd.DataFrame({"a": [4, 5], "b": [5, 6], "c": [6, 7]})