Imputing data with class imbalance

My dataset has class imbalances(multiclass logistic regression) with missing data. First, I split data using train test split(stratify=target column). Second, I use sklearn.Pipeline to standardize data using StandardScaler and use KNNimputer. Then I use imbalanced-learn.Pipeline for SMOTE and logistic regression. GrisearchCV pipe and params and fit on x_train,y_train. Sidenote, when would PCA be used in the process? How do you guys approach this dataset?