You are working with a dataset of penguins. Your goal is to build a complete machine learning pipeline that classifies penguin species using a K-Nearest Neighbors (KNN) model. The pipeline should handle categorical encoding, missing values, feature scaling, and parameter tuning.

1. Encode the target variable y using the LabelEncoder class.
2. Split the dataset into training and test sets using train_test_split() with test_size=0.33.
3. Create a ColumnTransformer named ct that applies a OneHotEncoder to the 'island' and 'sex' columns, leaving all other columns unchanged (remainder='passthrough').
4. Define a parameter grid param_grid that contains the following values for n_neighbors: [1, 3, 5, 7, 9, 12, 15, 20, 25], and include 'weights' ('uniform', 'distance') and 'p' (1, 2).
5. Create a GridSearchCV object using KNeighborsClassifier() as the estimator and param_grid as the parameter grid.
6. Build a pipeline that includes the following steps in order:
   • The ColumnTransformer (ct);
   • A SimpleImputer with the strategy set to 'most_frequent';
   • A StandardScaler for feature scaling;
   • The GridSearchCV object as the final step.
7. Train the pipeline on the training data (X_train, y_train) using the .fit() method.
8. Evaluate the model performance by printing the test score using .score(X_test, y_test).
9. Generate predictions on the test data and print the first 5 decoded class names using label_enc.inverse_transform().
10. Print the best estimator found by GridSearchCV using the .best_estimator_ attribute.