I added the feature importance function

Author: MaaTi-project

.idea/eis_toolkit.iml

@@ -76,3 +76,11 @@ class NonSquarePixelSizeException(Exception):
 
 class NumericValueSignException(Exception):
     """Exception error class for numeric value sign exception."""
+
+
+class InvalidModelException(Exception):
+    """Exception error class when model is invalid or null."""
+
+
+class InvalidDatasetException(Exception):
+    """Exception error class when the dataset is null."""

.idea/misc.xml

@@ -0,0 +1,48 @@
+import numpy as np
+import pandas
+import pandas as pd
+import sklearn.neural_network
+from sklearn.inspection import permutation_importance
+
+from eis_toolkit.exceptions import InvalidDatasetException
+
+
+def evaluate_feature_importance(
+    clf: sklearn.neural_network or sklearn.linear_model,
+    x_test: np.ndarray,
+    y_test: np.ndarray,
+    feature_names: list[str],
+    number_of_repetition: int = 50,
+    random_state: int = 0,
+) -> (pandas.DataFrame, dict):
+    """
+    Evaluate the feature importance of a sklearn classifier or linear model.
+
+    Parameters:
+        clf (Any sklearn nn model or lm model): Trained classifier.
+        x_test (np.ndarray): Testing feature data (X data need to be normalized / standardized).
+        y_test (np.ndarray): Testing target data.
+        feature_names (list): Names of the feature columns.
+        number_of_repetition (int): Number of iteration used when calculate feature importance (default 50).
+        random_state (int): random state for repeatability of results (Default 0).
+    Return:
+        feature_importance (pd.Dataframe): A dataframe composed by features name and Importance value
+        result (dict[object]): The resulted object with importance mean, importance std, and overall importance
+    Raise:
+        InvalidDatasetException: When the dataset is None.
+    """
+
+    if x_test is None or y_test is None:
+        raise InvalidDatasetException
+
+    result = permutation_importance(
+        clf, x_test, y_test.ravel(), n_repeats=number_of_repetition, random_state=random_state
+    )
+
+    feature_importance = pd.DataFrame({"Feature": feature_names, "Importance": result.importances_mean})
+
+    feature_importance["Importance"] = feature_importance["Importance"] * 100
+    feature_importance = feature_importance.sort_values(by="Importance", ascending=False)
+    # feature_importance['Importance'] = feature_importance['Importance'].apply(lambda x: '{:.6f}%'.format(x))
+
+    return feature_importance, result

eis_toolkit/exceptions.py

@@ -0,0 +1,69 @@
+import matplotlib.pyplot as plt
+import pandas as pd
+from sklearn.model_selection import train_test_split
+from sklearn.neural_network import MLPClassifier
+from sklearn.preprocessing import StandardScaler
+
+from eis_toolkit.feature_importance.feature_importance import evaluate_feature_importance
+
+# here I set to paths
+data_to_load = "PUT PATH TO X"
+label_to_load = "PUT PATH TO Y"
+
+if __name__ == "__main__":
+
+    feature_names = [
+        "Mag_TMI",
+        "Mag_AS",
+        "DRC135",
+        "DRC180",
+        "DRC45",
+        "DRC90",
+        "Mag_TD",
+        "HDTDR",
+        "Mag_Xdrv",
+        "mag_Ydrv",
+        "Mag_Zdrv",
+        "Pseu_Grv",
+        "Rd_U",
+        "Rd_TC",
+        "Rd_Th",
+        "Rd_K",
+        "EM_ratio",
+        "EM_Ap_rs",
+        "Em_Qd",
+        "EM_Inph",
+    ]
+
+    # first things first let s load data
+    X = pd.read_csv(f"{data_to_load}").to_numpy()
+    y = pd.read_csv(f"{label_to_load}").to_numpy()
+
+    # standardize the content
+    X = StandardScaler().fit_transform(X)
+
+    # now let s train a MLP classifier
+    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.25)
+
+    # we can train a MLP classifier
+    clf = MLPClassifier(solver="adam", alpha=0.001, hidden_layer_sizes=(16, 2), random_state=1)
+    clf.fit(X_train, y_train.ravel())
+
+    # we evaluate feature here
+    evaluated_feature_importance, dictionary_of_features = evaluate_feature_importance(
+        clf=clf, x_test=X_test, y_test=y_test, feature_names=feature_names, number_of_repetition=50, random_state=0
+    )
+
+    print(evaluated_feature_importance)
+
+    # how to create a chart from here
+    imp = pd.Series(dictionary_of_features.importances_mean * 100, index=feature_names).sort_values(ascending=True)
+    ax = imp.plot.barh()
+    ax.set_title("MLP Permutation Importance")
+    ax.figure.tight_layout()
+    plt.xlabel("Importance (%)")
+    plt.grid(axis="x", linestyle="--", alpha=0.6)
+    plt.ylabel("Feature")
+    for i, v in enumerate(imp):
+        ax.text(v, i, f"{v:.1f}", color="blue", fontweight="bold", fontsize=8)
+    plt.savefig("testing.png")