I want to create a pipeline to do preprocessing in both training features and target, then train the model. Dataset would be something like:

   v1  v2  target
0  1   a   yes
1  5   c   no
2  3   f   yes

I have a pipeline like

num_cols = ['v1']
cat_cols = ['v2']
clf = DecisionTreeClassifier()

num_transformer = Pipeline(steps=[
    ('impute', SimpleImputer(strategy='mean')),
    ('scale', MinMaxScaler())
])
cat_transformer = Pipeline(steps=[('impute', SimpleImputer(strategy='most_frequent'))])

col_trans = ColumnTransformer(transformers=[
    ('num_pipeline', num_transformer, num_cols),
    ('cat_pipeline', cat_transformer, cat_cols))

clf_pipeline = Pipeline(steps=[
    ('col_trans', col_trans),
    ('model', clf)
])

The idea is to Label Encode the target so it will be

target
1
0
1

If possible it would be interesting to also decode the predictions