You are given a dataset of penguins. Your goal is to build a machine learning pipeline that classifies penguin species using a K-Nearest Neighbors (KNN) model, while properly handling encoding, missing values, and parameter optimization.

1. Encode the target variable using LabelEncoder.
2. Split the dataset into training and test sets with test_size=0.33.
3. Create a ColumnTransformer (ct) that encodes only the 'island' and 'sex' columns using a suitable encoder for nominal data (OneHotEncoder) and keeps the other columns untouched.
4. Define a parameter grid (param_grid) that includes the following values for n_neighbors: [1, 3, 5, 7, 9, 12, 15, 20, 25].
5. Create a GridSearchCV object with KNeighborsClassifier as the base estimator and param_grid as its parameters.
6. Build a pipeline consisting of:
   • the ColumnTransformer (ct);
   • a SimpleImputer (strategy = 'most_frequent');
   • a StandardScaler;
   • and the GridSearchCV as the final step.
7. Train the pipeline using the training data (X_train, y_train).
8. Evaluate the model on the test data by printing its .score(X_test, y_test).
9. Predict on the test set and print the first 5 decoded predictions using label_enc.inverse_transform().
10. Finally, print the best estimator found by GridSearchCV.