Classify outlier in excel file

src/cluster_test/cluster.py

@@ -2,41 +2,39 @@ import numpy as np
 import pandas
 from sklearn.impute import SimpleImputer
 from sklearn.pipeline import make_pipeline
-from sklearn import tree
-from sklearn.preprocessing import OrdinalEncoder
-from sklearn.tree import export_text
+from sklearn.neighbors import LocalOutlierFactor
 
 NO_DATA: str = "NO DATA"
 
-def read_excel(path: str, target_column: str):
+def read_excel(path: str):
     data = pandas.read_excel(path)
-    X = data.drop(target_column, axis=1)
-    y = data[target_column]
-    feature_names = data.columns.drop(target_column)
+    X = data
+    feature_names = data.columns
 
     for feat in feature_names:
-        if feat == target_column:
-            continue
-
         if feat.startswith("Messpunkt"):
             # Convert to numerical value
             X[feat] = X[feat].replace(to_replace=NO_DATA, value=np.nan)
             X[feat] = X[feat].astype('float64')
         else:
             # Convert to categorical value
-            X[feat] = X[feat].astype('string')
+            X[feat] = X[feat].replace(to_replace=NO_DATA, value=None).astype('object')
 
-    return X, y, feature_names
+    return X
 
-def classify(X, y, feature_names=None):
-    clf = tree.DecisionTreeClassifier()
-    clf = clf.fit(X, y)
-    r = export_text(clf, feature_names=feature_names)
-    print(r)
+def classify(X, data_frame):
+    clf = LocalOutlierFactor(n_neighbors=3)
+    y_pred = clf.fit_predict(X)
+    #X_scores = clf.negative_outlier_factor_
+
+    #print(y_pred)
+    #print(X_scores)
+    return data_frame[y_pred < 0]
 
 if __name__ == '__main__':
-    X, y, feature_names = read_excel("tests/Beispiel Auswertung2.xlsx", "Motornummer")
-    pipe = make_pipeline(OrdinalEncoder(), SimpleImputer())
+    data_frame = read_excel("tests/Beispiel Auswertung2.xlsx")
+    X = pandas.get_dummies(data_frame) # OneHotEncode categorical values
+    pipe = make_pipeline(SimpleImputer(add_indicator=True))
     X = pipe.fit_transform(X)
-    print(X)
-    classify(X, y, feature_names)
+    outlier = classify(X, data_frame)
+    print(outlier)