[FAQ Bot] NEW: Why am I getting TypeError while creating OneHotEncoder Object? (#24)

* NEW: Why am I getting TypeError while creating OneHotEncoder Object?

* Update OneHotEncoder documentation for parameter change

---------

Co-authored-by: FAQ Bot <[email protected]>
Co-authored-by: Alexey Grigorev <[email protected]>

Author: 3 people

_questions/machine-learning-zoomcamp/module-3/019_548dcc8a3c_onehotencoder-sparse-output-deprecation.md

@@ -0,0 +1,27 @@
+---
+id: 548dcc8a3c
+question: Why am I getting TypeError while creating OneHotEncoder Object?
+sort_order: 19
+---
+
+In scikit-learn >= 1.2, `OneHotEncoder` no longer accepts the sparse parameter. It now uses the `sparse_output` parameter to control whether the output is sparse or dense.
+
+What to do:
+- For a dense array: use `OneHotEncoder(sparse_output=False)`.
+- For a sparse matrix: use `OneHotEncoder(sparse_output=True)`.
+
+If you previously wrote `ohe = OneHotEncoder(sparse=False)` and get a `TypeError`, replace it with `ohe = OneHotEncoder(sparse_output=False)`.
+
+Example:
+
+```python
+from sklearn.preprocessing import OneHotEncoder
+
+ohe = OneHotEncoder(sparse_output=False)
+X_train_cat = ohe.fit_transform(df_train[categorical_columns].values)
+```
+
+Notes:
+- A dense array will be returned if `sparse_output=False`; otherwise, you’ll get a scipy sparse matrix.
+- If you need a dense array from a sparse result, call `.toarray()` on the result.
+- This change ensures better memory efficiency and consistency in newer scikit-learn versions.