Generate recommendations using Apache Mahout in Azure HDInsight

Learn how to use the Apache Mahout machine learning library with Azure HDInsight to generate movie recommendations.

Mahout is a machine learning library for Apache Hadoop. Mahout contains algorithms for processing data, such as filtering, classification, and clustering. In this article, you use a recommendation engine to generate movie recommendations that are based on movies your friends have seen.

An Apache Hadoop cluster on HDInsight. See Get Started with HDInsight on Linux.

One of the functions that is provided by Mahout is a recommendation engine. This engine accepts data in the format of userID, itemId, and prefValue (the preference for the item). Mahout can then perform co-occurrence analysis to determine: users who have a preference for an item also have a preference for these other items. Mahout then determines users with like-item preferences, which can be used to make recommendations.

The following workflow is a simplified example that uses movie data:

• Co-occurrence: Joe, Alice, and Bob all liked Star Wars, The Empire Strikes Back, and Return of the Jedi. Mahout determines that users who like any one of these movies also like the other two.

• Co-occurrence: Bob and Alice also liked The Phantom Menace, Attack of the Clones, and Revenge of the Sith. Mahout determines that users who liked the previous three movies also like these three movies.

• Similarity recommendation: Because Joe liked the first three movies, Mahout looks at movies that others with similar preferences liked, but Joe hasn't watched (liked/rated). In this case, Mahout recommends The Phantom Menace, Attack of the Clones, and Revenge of the Sith.

Conveniently, GroupLens Research provides rating data for movies in a format that is compatible with Mahout. This data is available on your cluster's default storage at /HdiSamples/HdiSamples/MahoutMovieData.

There are two files, moviedb.txt and user-ratings.txt. The user-ratings.txt file is used during analysis. The moviedb.txt is used to provide user-friendly text information when viewing the results.

The data contained in user-ratings.txt has a structure of userID, movieID, userRating, and timestamp, which indicates how highly each user rated a movie. Here is an example of the data:

    196    242    3    881250949
    186    302    3    891717742
    22     377    1    878887116
    244    51     2    880606923
    166    346    1    886397596

1. Use ssh command to connect to your cluster. Edit the following command by replacing CLUSTERNAME with the name of your cluster, and then enter the command:

   ssh sshuser@CLUSTERNAME-ssh.azurehdinsight.cn
   

2. Use the following command to run the recommendation job:

   mahout recommenditembased -s SIMILARITY_COOCCURRENCE -i /HdiSamples/HdiSamples/MahoutMovieData/user-ratings.txt -o /example/data/mahoutout --tempDir /temp/mahouttemp
   

1. Once the job completes, use the following command to view the generated output:

   hdfs dfs -text /example/data/mahoutout/part-r-00000
   

   The output appears as follows:

   1    [234:5.0,347:5.0,237:5.0,47:5.0,282:5.0,275:5.0,88:5.0,515:5.0,514:5.0,121:5.0]
   2    [282:5.0,210:5.0,237:5.0,234:5.0,347:5.0,121:5.0,258:5.0,515:5.0,462:5.0,79:5.0]
   3    [284:5.0,285:4.828125,508:4.7543354,845:4.75,319:4.705128,124:4.7045455,150:4.6938777,311:4.6769233,248:4.65625,272:4.649266]
   4    [690:5.0,12:5.0,234:5.0,275:5.0,121:5.0,255:5.0,237:5.0,895:5.0,282:5.0,117:5.0]
   

   The first column is the userID. The values contained in '[' and ']' are movieId:recommendationScore.

2. You can use the output, along with the moviedb.txt, to provide more information on the recommendations. First, copy the files locally using the following commands:

   hdfs dfs -get /example/data/mahoutout/part-r-00000 recommendations.txt
   hdfs dfs -get /HdiSamples/HdiSamples/MahoutMovieData/* .
   

   This command copies the output data to a file named recommendations.txt in the current directory, along with the movie data files.

3. Use the following command to create a Python script that looks up movie names for the data in the recommendations output:

   nano show_recommendations.py
   

   When the editor opens, use the following text as the contents of the file:

   #!/usr/bin/env python
   
   import sys
   
   if len(sys.argv) != 5:
        print "Arguments: userId userDataFilename movieFilename recommendationFilename"
        sys.exit(1)
   
   userId, userDataFilename, movieFilename, recommendationFilename = sys.argv[1:]
   
   print "Reading Movies Descriptions"
   movieFile = open(movieFilename)
   movieById = {}
   for line in movieFile:
       tokens = line.split("|")
       movieById[tokens[0]] = tokens[1:]
   movieFile.close()
   
   print "Reading Rated Movies"
   userDataFile = open(userDataFilename)
   ratedMovieIds = []
   for line in userDataFile:
       tokens = line.split("\t")
       if tokens[0] == userId:
           ratedMovieIds.append((tokens[1],tokens[2]))
   userDataFile.close()
   
   print "Reading Recommendations"
   recommendationFile = open(recommendationFilename)
   recommendations = []
   for line in recommendationFile:
       tokens = line.split("\t")
       if tokens[0] == userId:
           movieIdAndScores = tokens[1].strip("[]\n").split(",")
           recommendations = [ movieIdAndScore.split(":") for movieIdAndScore in movieIdAndScores ]
           break
   recommendationFile.close()
   
   print "Rated Movies"
   print "------------------------"
   for movieId, rating in ratedMovieIds:
       print "%s, rating=%s" % (movieById[movieId][0], rating)
   print "------------------------"
   
   print "Recommended Movies"
   print "------------------------"
   for movieId, score in recommendations:
       print "%s, score=%s" % (movieById[movieId][0], score)
   print "------------------------"
   

   Press Ctrl-X, Y, and finally Enter to save the data.

4. Run the Python script. The following command assumes you are in the directory where all the files were downloaded:

   python show_recommendations.py 4 user-ratings.txt moviedb.txt recommendations.txt
   

   This command looks at the recommendations generated for user ID 4.

   • The user-ratings.txt file is used to retrieve movies that have been rated.

   • The moviedb.txt file is used to retrieve the names of the movies.

   • The recommendations.txt is used to retrieve the movie recommendations for this user.

     The output from this command is similar to the following text:

     Seven Years in Tibet (1997), score=5.0
     Indiana Jones and the Last Crusade (1989), score=5.0
     Jaws (1975), score=5.0
     Sense and Sensibility (1995), score=5.0
     Independence Day (ID4) (1996), score=5.0
     My Best Friend's Wedding (1997), score=5.0
     Jerry Maguire (1996), score=5.0
     Scream 2 (1997), score=5.0
     Time to Kill, A (1996), score=5.0
     

Mahout jobs don't remove temporary data that is created while processing the job. The --tempDir parameter is specified in the example job to isolate the temporary files into a specific path for easy deletion. To remove the temp files, use the following command:

hdfs dfs -rm -f -r /temp/mahouttemp

Now that you've learned how to use Mahout, discover other ways of working with data on HDInsight:

• Apache Hive with HDInsight
• MapReduce with HDInsight