If we cut our clusters off at the dotted line above, how many elements (words) are in the largest cluster and the smallest cluster?
A. largest: 3, smallest: 1
B. largest: 4, smallest: 1
C. largest: 4, smallest: 3
D. largest: 4, smallest: 4
If we cut our clusters off at the dotted line above, how many clusters will we have?
A. 1
B. 2
C. 3
D. 4
5. Looking at the options they’re given, the board members choose to go with a supervised model with lead role as data. You become outraged. "How can you not include movie length? It’s incredibly important! Who watches a 3 hour long movie --" Your fellow data scientist interrupts you. "Yeah, I agree, but look at these initial results. You see, if we remove movie length, ..." What can your colleague (correctly) say to convince you? Check all that apply.
A. "we can reduce inter-cluster dissimilarity."
B. "we can reduce intra-cluster dissimilarity."
C. "the model starts to work. It doesn’t work otherwise."
D. "we can consume less memory, and the results look almost the same."
Now that the company has some idea about how to use the data, it’s time to design a classifier. Our classifier will be very simple: given a movie and a user, it will classify the movie as either "Good" or "Bad" for this user.4. Assume all the users of the company have a very simple rule in their movie taste: they like it if Tom Cruise has the lead role. Any other data is mostly irrelevant. However, no one in the company knows about this fact. Which of the following clustering models might be able to detect this rule? Check all that apply.
A. Supervised (label: rating), with data: Director, language, genre
B. Supervised (label: rating), with data: Movie length, lead role, director
C. Unsupervised, with data: Lead role, movie length, rating
D. Unsupervised, with data: Lead role, genre, director
E. Unsupervised, with data: Number of ratings, lead role