Use the Vader analyzer to conduct sentiment analysis on each SMS$.$ Use all

four sentiment scores $($pos$,$ neg, neu, and compound$)$ as predictors and apply the

decision tree method $($random$\_$state$=50)$ to build a prediction model for spam

emails. Report its Precision, Recall and F$1$ on testing data $($training: $70\%,$

testing:$30\%,$ random$\_$state$=50.$ Use this same setting for the following model

trainings and evaluations in b$)$ and c$).).$ Which sentiment score is most important

in this prediction model?