Problem $2$: Preprocessing

We will now create a pipeline to process our dataset. This processing will involve indexing and encoding the

categorical features and then combining all of the features into vectors.

Create lists named num$\_$features and cat$\_$features to store the names of columns representing

numerical and categorical features. The numerical features are age, avg$\_$glucose$\_$level, and bmi. All

other features are categorical.

Create lists named ix$\_$features and vec$\_$features to store the names of the integer$-$encoded categorical

columns and the one$-$hot encoded categorical columns $($respectively$).$

Create a StringIndexer object that uses the columns named in the list cat$\_$features to create the

columns named in the list ix$\_$features.

Create a OneHotEncoder object that uses the integer$-$encoded features named in ix$\_$features to create

the one$-$hot encoded categorical features named in vec$\_$features. Do not drop the last columns.

Create a VectorAssembler object that combines the numerical features and the one$-$hot encoded vectors

for the categorical features. The combined column should be named features.

We will now create a pipeline from the stages above and will apply this to our data.

Create a pipeline consisting of the StringIndexer, OneHotEncoder, and VectorAssembler objects. Fit

this pipeline to the stroke$\_$df DataFrame, and then apply the fitted pipeline to stroke$\_$df$.$ Store the

processed DataFrame in a variable named train.

Persist the train DataFrame. Then display the first $10$ rows of the features and stroke columns of

train, setting truncate$=$False.