Critical note, the StandarScalar was performed using all the data, aka the train and test. This is test data leakage. Should be performed by fitting the training and then transforming the test separately. Other than that it looks great!