Merge pull request #27 from mmschlk/development

Development

requirements.txt

@@ -4,3 +4,4 @@ matplotlib
 colour
 networkx
 scikit-learn
+pandas

setup.py

@@ -18,7 +18,7 @@
 with io.open(os.path.join(work_directory, "README.md"), encoding="utf-8") as f:
     long_description = "\n" + f.read()
 
-base_packages = ["numpy", "scipy"]
+base_packages = ["numpy", "scipy", "pandas"]
 
 plotting_packages = ["matplotlib", "colour", "networkx"]
 

shapiq/__init__.py

@@ -30,6 +30,8 @@
     split_subsets_budget,
 )
 
+from .datasets import load_bike
+
 __all__ = [
     # version
     "__version__",
@@ -51,4 +53,6 @@
     "split_subsets_budget",
     "get_conditional_sample_weights",
     "get_parent_array",
+    # datasets
+    "load_bike",
 ]

shapiq/datasets/__init__.py

@@ -0,0 +1,5 @@
+"""This module contains small datasets for testing and examples."""
+
+from ._all import load_bike
+
+__all__ = ["load_bike"]

shapiq/datasets/_all.py

@@ -0,0 +1,38 @@
+import os
+import pandas as pd
+
+
+GITHUB_DATA_URL = "https://github.com/mmschlk/shapiq/raw/main/data/"
+
+
+def load_bike() -> pd.DataFrame:
+    """Load the bike-sharing dataset from a Kaggle competition.
+
+    Original source: https://www.kaggle.com/c/bike-sharing-demand
+
+    Note:
+        The function and preprocessing is taken from the `sage` package.
+
+    Returns:
+        The bike-sharing dataset as a pandas DataFrame.
+    """
+    data = pd.read_csv(os.path.join(GITHUB_DATA_URL, "bike.csv"))
+    columns = data.columns.tolist()
+
+    # Split and remove datetime column.
+    data["datetime"] = pd.to_datetime(data["datetime"])
+    data["year"] = data["datetime"].dt.year
+    data["month"] = data["datetime"].dt.month
+    data["day"] = data["datetime"].dt.day
+    data["hour"] = data["datetime"].dt.hour
+    data = data.drop("datetime", axis=1)
+
+    # Reorder and rename columns.
+    data = data[["year", "month", "day", "hour"] + columns[1:]]
+    data.columns = list(map(str.title, data.columns))
+
+    return data
+
+
+if __name__ == "__main__":
+    print(load_bike())

shapiq/explainer/__init__.py

@@ -2,7 +2,6 @@
 
 
 from .interaction import InteractionExplainer
+from .tree import TreeExplainer
 
-__all__ = [
-    "InteractionExplainer",
-]
+__all__ = ["InteractionExplainer", "TreeExplainer"]

shapiq/explainer/imputer/marginal_imputer.py

@@ -39,7 +39,7 @@ def __init__(
         background_data: np.ndarray,
         x_explain: Optional[np.ndarray] = None,
         sample_replacements: bool = False,
-        sample_size: int = 5,
+        sample_size: int = 1,
         categorical_features: list[int] = None,
         random_state: Optional[int] = None,
     ) -> None:

shapiq/explainer/tree.py

@@ -0,0 +1,20 @@
+"""This module contains the TreeSHAP-IQ explainer for computing exact any order Shapley interactions
+for trees and tree ensembles."""
+import numpy as np
+
+from approximator._base import InteractionValues
+from explainer._base import Explainer
+
+
+__all__ = ["TreeExplainer"]
+
+
+class TreeExplainer(Explainer):
+    def __init__(self) -> None:
+        raise NotImplementedError(
+            "The TreeExplainer is not yet implemented. An initial version can be found here: "
+            "'https://github.com/mmschlk/TreeSHAP-IQ'."
+        )
+
+    def explain(self, x_explain: np.ndarray) -> InteractionValues:
+        pass

shapiq/plot/network.py

@@ -7,6 +7,8 @@
 import numpy as np
 from matplotlib import pyplot as plt
 from PIL import Image
+
+from approximator._base import InteractionValues
 from utils import powerset
 
 from ._config import BLUE, RED
@@ -23,12 +25,6 @@ def _get_color(value: float) -> str:
     return BLUE.hex
 
 
-def _min_max_normalization(value: float, min_value: float, max_value: float) -> float:
-    """Normalizes the value between min and max"""
-    size = (value - min_value) / (max_value - min_value)
-    return size
-
-
 def _add_weight_to_edges_in_graph(
     graph: nx.Graph,
     first_order_values: np.ndarray,
@@ -57,7 +53,7 @@ def _add_weight_to_edges_in_graph(
         graph.nodes[node]["edgecolors"] = color
 
     for edge in powerset(range(n_features), min_size=2, max_size=2):
-        weight: float = second_order_values[edge]
+        weight: float = float(second_order_values[edge])
         color = _get_color(weight)
         # scale weight between min and max edge value
         size = abs(weight) / all_range
@@ -134,9 +130,10 @@ def _add_legend_to_axis(axis: plt.Axes) -> None:
 
 
 def network_plot(
+    *,
+    interaction_values: InteractionValues,
     first_order_values: np.ndarray[float],
     second_order_values: np.ndarray[float],
-    *,
     feature_names: Optional[list[Any]] = None,
     feature_image_patches: Optional[dict[int, Image.Image]] = None,
     feature_image_patches_size: Optional[Union[float, dict[int, float]]] = 0.2,
@@ -156,6 +153,7 @@ def network_plot(
         :align: center
 
     Args:
+        interaction_values: The interaction values as an interaction object.
         first_order_values: The first order n-SII values of shape (n_features,).
         second_order_values: The second order n-SII values of shape (n_features, n_features). The
             diagonal values are ignored. Only the upper triangular values are used.

shapiq/utils/modules.py

@@ -0,0 +1,57 @@
+import sys
+from typing import Any, Union
+
+
+def safe_isinstance(obj: Any, class_path_str: Union[str, list[str], tuple[str]]) -> bool:
+    """
+    Acts as a safe version of isinstance without having to explicitly import packages which may not
+    exist in the user's environment. Checks if obj is an instance of type specified by
+    class_path_str.
+
+    Note:
+        This function was directly taken from the `shap` repository.
+
+    Args:
+        obj: Some object you want to test against
+        class_path_str: A string or list of strings specifying full class paths Example:
+            `sklearn.ensemble.RandomForestRegressor`
+
+    Returns:
+            True if isinstance is true and the package exists, False otherwise
+    """
+    if isinstance(class_path_str, str):
+        class_path_strs = [class_path_str]
+    elif isinstance(class_path_str, list) or isinstance(class_path_str, tuple):
+        class_path_strs = class_path_str
+    else:
+        class_path_strs = [""]
+
+    # try each module path in order
+    for class_path_str in class_path_strs:
+        if "." not in class_path_str:
+            raise ValueError(
+                "class_path_str must be a string or list of strings specifying a full \
+                module path to a class. Eg, 'sklearn.ensemble.RandomForestRegressor'"
+            )
+
+        # Splits on last occurrence of "."
+        module_name, class_name = class_path_str.rsplit(".", 1)
+
+        # here we don't check further if the model is not imported, since we shouldn't have
+        # an object of that types passed to us if the model the type is from has never been
+        # imported. (and we don't want to import lots of new modules for no reason)
+        if module_name not in sys.modules:
+            continue
+
+        module = sys.modules[module_name]
+
+        # Get class
+        _class = getattr(module, class_name, None)
+
+        if _class is None:
+            continue
+
+        if isinstance(obj, _class):
+            return True
+
+    return False

tests/tests_explainer/test_explainer_tree.py

@@ -0,0 +1,9 @@
+"""This module contains all tests for the TreeExplainer class of the shapiq package."""
+import pytest
+
+from shapiq.explainer import TreeExplainer
+
+
+def test_init():
+    with pytest.raises(NotImplementedError):
+        explainer = TreeExplainer()