Update custom_definitions.py

custom_definitions.py

@@ -19,3 +19,41 @@
 # Your code
 #==============Your Name=====================
 #Dont remove the following snippet and follow the same
+
+from sklearn.base import BaseEstimator, TransformerMixin
+import numpy as np
+import pandas as pd
+from scipy.stats import zscore
+
+class lsZScoreOutlierRemover(BaseEstimator, TransformerMixin):
+    def _init_(self, threshold=3.0):
+        self.threshold = threshold
+
+    def fit(self, Xls, yls=None):
+        return self  # stateless
+
+    def transform(self, Xls, yls=None):
+        Xls = Xls.copy()
+
+        # Feature engineering first (to include them in outlier handling)
+        Xls["Proximity_Temp"] = Xls["Proximity to Star"] / (Xls["Surface Temperature"] + 1e-5)
+        Xls["Surface Density"] = Xls["Mineral Abundance"] * (Xls["Surface Temperature"] + 1e-5)
+        Xls["log_Density"] = np.log1p(Xls["Atmospheric Density"] - Xls["Atmospheric Density"].min() + 1)
+
+        # Select numeric columns
+        num_colsls = Xls.select_dtypes(include=[np.number]).columns.tolist()
+
+        # Ensure all numeric data is float (important for NaNs)
+        Xls[num_colsls] = Xls[num_colsls].astype(float)
+
+        # Compute Z-scores
+        z_scoresls = np.abs(zscore(Xls[num_colsls], nan_policy='omit'))
+
+        # Handle 1D zscore output (edge case)
+        if z_scoresls.ndim == 1:
+            z_scoresls = z_scoresls[:, np.newaxis]
+
+        # Replace outliers with NaN
+        Xls[num_colsls] = Xls[num_colsls].where(z_scoresls < self.threshold, np.nan)
+
+        return Xls