one-hots

Before training the classifier, I converted the categorical columns into one-hots to ensure the model handled discrete features correctly.