This is the solution page for Lab 7: Analyze ratings with Crunch.

Steps

1. Build and run the Crunch job

cd /home/cloudera/ratings-crunch
git checkout bimodal
mvn clean package
mvn kite:run-tool

2. Look at the code

Read through the source at src/main/java/org/kitesdk/examples/movies/AnalyzeRatings.java.

Good questions earn prizes.

3. Find an interesting movie

You can use a SQL join query to view the title and the ratings histogram at once, using Hive.

Using beeline will produce pretty output. Start beeline in embedded mode:

beeline -u jdbc:hive2://

select m.title, h.histogram from movies as m, ratings_histograms as h where m.id = h.movie_id;

+----------------------------------------------------------------------------------+--------------------+--+
|                                     m.title                                      |    h.histogram     |
+----------------------------------------------------------------------------------+--------------------+--+
| ...                                                                              |                    |
| Bio-Dome (1996)                                                                  | [16,5,8,1,1]       |
| ...                                                                              |                    |
+----------------------------------------------------------------------------------+--------------------+--+
50 rows selected (42.249 seconds)

That can’t be right. Bio-Dome is a classic!

Next