Sample scripts to analyze taxi data on Amazon AWS

1. Create an Amazon EMR cluster with the following configuration (the bootstrap action is very important -- please pay attention to that):

    * Termination protection: Yes
    * Logging: Enabled (remember to input your S3 bucket to store log file)
    * Hadoop distribution: Amazon AMI 3.3.1
    * Bootstrap action: This is a very important step because the sample scripts 
    make use of python rtree library, but Amazon AMI 3.3.1 does not have rtree installed.
    Click 'Add bootstrap action' -> Custom action -> Configure and add -> 
    Put the following in 'S3 location': s3://mda2014/rtree.sh
    * Don't add any step at this point
    * Cluster Auto-terminate: No
   

2. Clone this repository and upload the neighborhoods and yearplot scripts to your bucket on S3. For example:

    * neighborhoods: s3://mda2014/neighborhoods
    * yearplot: s3://mda2014/yearplot
   

3. To run neighborhoods script: Add the following streaming step to your cluster with the following information:

    Replace mda2014 with your bucket name, except in Input
    * Mapper: s3://mda2014/neighborhoods/mapper.py
    * Reducer: s3://mda2014/neighborhoods/reducer.py
    * Input: s3://mda2014/taxi/trip/
    * Output: s3://mda2014/output1
    * Arguments: -D mapred.reduce.tasks=1 -files s3://mda2014/neighborhoods/mapper.py,s3://mda2014/neighborhoods/reducer.py,s3://mda2014/neighborhoods/shapefile.py,s3://mda2014/neighborhoods/ZillowNeighborhoods-NY.shp,s3://mda2014/neighborhoods/ZillowNeighborhoods-NY.prj,s3://mda2014/neighborhoods/ZillowNeighborhoods-NY.shp.xml,s3://mda2014/neighborhoods/ZillowNeighborhoods-NY.shx,s3://mda2014/neighborhoods/ZillowNeighborhoods-NY.dbf
   

   Wait for finish, then download and merge all output into one file called output.txt

   To generate plot, execute:

    python plot_results.py output.txt <location_of_output_plot>
   

4. To run yearplot script: Add the following streaming step to your cluster with the following information:

    Replace mda2014 with your bucket name, except in Input
    * Mapper: s3://mda2014/yearplot/mapper.py
    * Reducer: s3://mda2014/yearplot/reducer.py
    * Input: s3://mda2014/taxi/trip/
    * Output: s3://mda2014/output2
    * Arguments: -D mapred.reduce.tasks=1
   

   Wait for finish, then download and merge all output into one file called output.txt

   To generate plot, execute:

    python plot_results.py output.txt <location_of_output_plot>
   

5. Remember to terminate cluster after use.

Huy T. Vo

Tuan-Anh Hoang-Vu