/aws-dynamo-weka

Upload data to Amazon AWS DynamoDB to measure the time taken. Then use Weka to create clusters of data.

Primary LanguageJava

aws-dynamo-weka

Upload data to Amazon AWS DynamoDB instance to measure the time taken. Then use Weka to create clusters of data.

###Steps:

  1. Created AWS DynamoDB instance on AWS account.
  2. Created a table on DynamoDB on the above instance and set PersonID as the primary key.
  3. Installed AWS SDK in Eclipse to access AWS services from Java code.
  4. Downloaded titanic survival data from http://www.cs.toronto.edu/~delve/data/titanic/desc.html
  5. Uploaded data from titanic.data file to DynamoDB using LoadDataToDynamo.java.
  6. Downloaded Weka API (3.6.11) from http://sourceforge.net/projects/weka/ and configured Eclipse project to use weka.jar.
  7. To create clusters and visualize titanic data run WekaRunner.java. This class does following things:
    • Create Weka Instances after download data from DynamoDB using AWS API.
    • These weka instances are then passed to a clusterer. weka.clusterers.EM class is used for clustering.
    • Using VisualizePanel class in weka, created a cluster visualization of the titanic data.
    • Timings: Total Time Taken [DynamoDB Scan]: 1718 ms Total Time Taken [WekaRunner]: 24356 ms