A video streaming company that you work for is interested in predicting the most popular videos among young viewers a category that certain advertisers are interested in One way to go about it would be to learn what genres tend to be favored by this group However you suspect that the standard definitions of genre classes may be noisy and imperfect and genres are not very relevant to understanding the taste profile of your people You think that perhaps some completely different characteristics when considered together make clusters that describe user preferences You decide not to assume what exact groupings to look for Instead you build a clustering model for 5 groups that largely share viewership Which stage of the ML process does this scenario describe Business understanding Data understanding Data preparation Modeling