HackReduce1

http://www.hackreduce.org

Prerequisites

• Java 1.6
• Build tool (can use either one):
  • Gradle (http://www.gradle.org/installation.html)
  • Ant
• Git

Datasets

https://github.com/hoppertravel/HackReduce/wiki/Datasets

Take a look at the datasets/ folder to see samples subsets of these datasets.

Run an example job locally

1. git clone git://github.com/hoppertravel/HackReduce.git (you should occasionally run "git pull" from within the project directory to update your code)

2. cd HackReduce

3. Build the project depending on what tool you have installed:

   Gradle:

   OR

   Ant:

4. Try running an example from the list below

Examples

Run any of the following commands in your CLI, and after the job's completed, check the /tmp/* folder for the output.

Bixi:

$ java -classpath ".:build/libs/HackReduce-0.2.jar:lib/*" org.hackreduce.examples.bixi.RecordCounter datasets/bixi /tmp/bixi_recordcounts

NASDAQ:

$ java -classpath ".:build/libs/HackReduce-0.2.jar:lib/*" org.hackreduce.examples.stockexchange.HighestDividend datasets/nasdaq/dividends /tmp/nasdaq_dividends
$ java -classpath ".:build/libs/HackReduce-0.2.jar:lib/*" org.hackreduce.examples.stockexchange.MarketCapitalization datasets/nasdaq/daily_prices /tmp/nasdaq_marketcaps
$ java -classpath ".:build/libs/HackReduce-0.2.jar:lib/*" org.hackreduce.examples.stockexchange.RecordCounter datasets/nasdaq/daily_prices /tmp/nasdaq_recordcounts

NYSE:

$ java -classpath ".:build/libs/HackReduce-0.2.jar:lib/*" org.hackreduce.examples.stockexchange.HighestDividend datasets/nyse/dividends /tmp/nyse_dividends
$ java -classpath ".:build/libs/HackReduce-0.2.jar:lib/*" org.hackreduce.examples.stockexchange.MarketCapitalization datasets/nyse/daily_prices /tmp/nyse_marketcaps
$ java -classpath ".:build/libs/HackReduce-0.2.jar:lib/*" org.hackreduce.examples.stockexchange.RecordCounter datasets/nyse/daily_prices /tmp/nyse_recordcounts

Flights:

$ java -classpath ".:build/libs/HackReduce-0.2.jar:lib/*" org.hackreduce.examples.flights.RecordCounter datasets/flights /tmp/flights_recordcounts

Wikipedia:

$ java -classpath ".:build/libs/HackReduce-0.2.jar:lib/*" org.hackreduce.examples.wikipedia.RecordCounter datasets/wikipedia /tmp/wikipedia_recordcounts

Google 1gram:

$ java -classpath ".:build/libs/HackReduce-0.2.jar:lib/*" org.hackreduce.examples.ngram.one_gram.RecordCounter datasets/ngram/1gram /tmp/1gram_recordcounts

Google 2gram:

$ java -classpath ".:build/libs/HackReduce-0.2.jar:lib/*" org.hackreduce.examples.ngram.two_gram.RecordCounter datasets/ngram/2gram /tmp/2gram_recordcounts

MSD:

$ java -classpath ".:build/libs/HackReduce-0.2.jar:lib/*" org.hackreduce.examples.msd.RecordCounter datasets/msd /tmp/msd_recordcounts

Note: The jobs are made for the specific datasets, so pairing them up properly is important. The second argument (/tmp/*) is just a made up output path for the results of the job, and can be modified to anything you want.

Streaming example

• Python

    $ java -classpath ".:lib/*" org.apache.hadoop.streaming.HadoopStreaming -input datasets/nasdaq/daily_prices/ -output /tmp/py_streaming_count -mapper streaming/nasdaq_counter.py -reducer aggregate
  

• Ruby

    $ java -classpath ".:lib/*" org.apache.hadoop.streaming.HadoopStreaming -input datasets/nasdaq/daily_prices/ -output /tmp/rb_streaming_count -mapper streaming/nasdaq_counter.rb -reducer aggregate
  

-reducer aggregate is a built-in function for streaming.

Running on a Hadoop cluster

If you'd like to set up an actual Hadoop cluster (single or multi node), then follow the instructions on the Hadoop wiki: http://wiki.apache.org/hadoop/QuickStart

If you're running the HackReduce virtual image, or you're trying to run the example job against an actual install of Hadoop, the process should be similar. However, you'll need to upload the datasets folder into HDFS (if you're running it), or just make a note of the path.

Example:

$ bin/hadoop jar <path_to_hackreduce_jar>/HackReduce-0.2.jar org.hackreduce.examples.stockexchange.RecordCounter <path to>/datasets/nyse/daily_prices /tmp/nyse_recordcounts

Setting up for development

Gradle (recommended):

We recommend using Gradle for easy set up of the project in Eclipse, Idea, or other IDEs through Gradle plugins (http://www.gradle.org/standard_plugins.html). To use it, simply run one of the following:

$ gradle eclipse

$ gradle idea

Then import the project into your IDE of choice. This will download all of the dependencies (including sources) and create the necessary project files.

Manual setup:

You can also bring in the project manually into your IDE and then include all the *.jar files from the lib folder of the project.