So i believe I've found why most people have been getting a difference between their output and the expected output. The solution takes a unique of the list on rather than clothing id, department id and class id. Since we have a jacket class and outwear class version for clothing id 1119 the unique tool will lose the outwear class version. When joining this back onto our initial flow it will change the only row with a positive feedback count for item 1119 greater than or equal to 10 from an outwear class to a jacket class and thus placing it in the wrong group when we come to summarise and sort the output.
My solution:
Issues with the weekly challenge solution pinned to the thread: