Titanic

The classic Titanic data science project using Spark and AWS EMR.

Please refer to ./titanic.ipynb for the code and the wiki for the report.

Instructions: Setup Local Development Environment

• Download spark-2.1.0-bin-hadoop2.7

• Install plot.ly using conda in CLI (or Anaconda Navigator in GUI) :

    conda install plotly
  

• Run Jupyter in a pyspark shell: https://spark.apache.org/docs/latest/programming-guide.html#using-the-shell

    cd ./titanic
    PYSPARK_DRIVER_PYTHON="jupyter" PYSPARK_DRIVER_PYTHON_OPTS="notebook" "~/Downloads/spark-2.1.0-bin-hadoop2.7/bin/pyspark"
  

Instructions: AWS EMR

• Accessing the Spark Web UIs: http://docs.aws.amazon.com/emr/latest/ReleaseGuide/emr-spark-history.html

1. Set up SSH Tunnel
2. Configure SOCKS Proxy in Browser (using Foxy Proxy)
3. Test the proxy: http://master-public-dns-name/
4. Accesss the Web Interfaces:
   • Zepplin (Notebook): 8890
   • Spark (Cluster log): 18080
   • Ganglia (Monitoring): */ganglia/
   • Hadoop (MapReduce): 8088/cluster

• Install Git on AWS EMR: sudo yum install git-all

Reference

• Titatnic tutorial in Python: https://github.com/KaggleCityMo/titanic-tutorial/blob/master/titanic_tutorial_spark16.ipynb
• Spark MLlib Guide: http://spark.apache.org/docs/latest/ml-guide.html
• Spark SQL, DataFrames and Datasets Guide: http://spark.apache.org/docs/latest/sql-programming-guide.html
• Spark Python API Reference: http://spark.apache.org/docs/latest/api/python/index.html
• Spark Mlib Evaluation Metrics: https://spark.apache.org/docs/latest/mllib-evaluation-metrics.html
• Spark on AWS EMR: http://docs.aws.amazon.com/emr/latest/ReleaseGuide/emr-spark-launch.html
• AWS EMR - Overview: https://docs.aws.amazon.com/emr/latest/ManagementGuide/emr-overview.html
• AWS EMR - Upload Data to Amazon S3: http://docs.aws.amazon.com/emr/latest/ManagementGuide/emr-plan-upload-s3.html

Troubleshooting

• Unable to extract element from Vectors in DataFrame: Using User-defined Functions (UDF) http://stackoverflow.com/questions/37311688/how-to-split-column-of-vectors-into-two-columns

• Null value in test.csv: I154 is empty

• AWS EMR "Terminated with errorsFailed to provision ec2 instances because 'The requested instance profile EMR_AutoScaling_DefaultRole is invalid'": Removed EMR_AutoScaling_DefaultRole https://aws.amazon.com/premiumsupport/knowledge-center/emr-default-role-invalid/

• No output from ssh -i ~/aws_root.pem -ND xxxx: It's expected due to -N (no command)

• ​