Agg query with filter runs slower than without filter

SELECT COUNT(*) from reviews

where stars > 4

group by stars;

SELECT COUNT(*) from reviews

group by stars;

CREATE BUCKET YELP WITH \{"name":"YELP"}; 

CREATE SHADOW DATASET reviews ON YELP WHERE `type` = "review";

connect BUCKET YELP;

create index r_user_id on reviews(user_id:STRING);

create index r_stars on review(stars:STRING);