Use the concepts and sentiment analysis as additional inputs in a decision tree by aggregating the reviews and merging them with the main Airbnb listings data set. Assess whether adding these new input fields can better predict membership in the Perfect_Group than without them.
Although it is not required, you may also want to apply the text analytics methods to the text fields like description, neighbourhood_overview, house_rules, host_about, etc. in the main Airbnb listings data set. Then you would assess whether there is something about the way the hosts write about these that affects the scores they receive.