PredictionIO-Jokes-demo

Tech demo of using PredictionIO, R and the Jester Collaborative Filtering dataset.

Note: this is a tech demo and no extensive tutorial. Some basic knowledge of PredictionIO and R are necessary.

More information about PredictionIO and the Jester dataset:

• PredictionIO: http://prediction.io/
• Jester dataset: http://eigentaste.berkeley.edu/dataset/

Dependencies

This tech demo depends on a running instance of PredictionIO v0.7.3. Furthermore, the following R libraries are being used:

• httr
• jsonlite
• stringr

Usage

• Make sure PredictionIO is running (bin/start-all.sh)
• Create an app, save the app key.
• In the created app, add an Item Recommendation Engine and save the name of the engine.
• In predictionio.R, update the following variables: appkey, url, itemrecEngine.
• In R, set your current working directory to the location of this repository
• Make sure the file 'jester-data-1.csv' is in your current working directory
• Open and source import.R, this starts the import process
• When the data has been imported, start the training process of the alogithm of the Item Recommendation Engine in PredictionIO.
• When the training process has been completed, open demo.R to see how to get recommended items from PredictionIO.

Example demo output

# Get the top 5 recommendations for user 2
top5rec <- getTopNRecommendation(2, 5)
# [1] 77 76 67 43 34

jokes[77]
#                                                                           Joke 77 
# "If pro- is the opposite of con- then congress must be the opposite of progress."