v1.1, Nov 16, Fix a bug in "Entrace.scala" v1.0, Nov 13, Initial version
In this phase, you are required to do spatial hot spot analysis. In particular, you need to complete two different hot spot analysis tasks
This task will needs to perform a range join operation on a rectangle datasets and a point dataset. For each rectangle, the number of points located within the rectangle will be obtained. The hotter rectangle means that it include more points. So this task is to calculate the hotness of all the rectangles.
This task will focus on applying spatial statistics to spatio-temporal big data in order to identify statistically significant spatial hot spots using Apache Spark. The topic of this task is from ACM SIGSPATIAL GISCUP 2016.
The Problem Definition page is here: http://sigspatial2016.sigspatial.org/giscup2016/problem
The Submit Format page is here: http://sigspatial2016.sigspatial.org/giscup2016/submit
As stated in the Problem Definition page, in this task, you are asked to implement a Spark program to calculate the Getis-Ord statistic of NYC Taxi Trip datasets. We call it "Hot cell analysis"
To reduce the computation power need，we made the following changes:
- The input will be a monthly taxi trip dataset from 2009 - 2012. For example, "yellow_tripdata_2009-01_point.csv", "yellow_tripdata_2010-02_point.csv".
- Each cell unit size is 0.01 * 0.01 in terms of latitude and longitude degrees.
- You should use 1 day as the Time Step size. The first day of a month is step 1. Every month has 31 days.
- You only need to consider Pick-up Location.
- We don't use Jaccard similarity to check your answer. However, you don't need to worry about how to decide the cell coordinates because the code template generated cell coordinates. You just need to write the rest of the task.
- Output path (Mandatory)
- Task name: "hotzoneanalysis" or "hotcellanalysis"
- Task parameters: (1) Hot zone (2 parameters): nyc taxi data path, zone path(2) Hot cell (1 parameter): nyc taxi data path
Example
test/output hotzoneanalysis src/resources/point-hotzone.csv src/resources/zone-hotzone.csv hotcellanalysis src/resources/yellow_trip_sample_100000.csv
Note:
- The number/order of tasks do not matter.
- But, the first 7 of our final test cases will be hot zone analysis, the last 8 will be hot cell analysis.
The main function/entrace is "cse512.Entrance" scala file.
-
Point data: the input point dataset is the pickup point of New York Taxi trip datasets. The data format of this phase is the original format of NYC taxi trip which is different from the previous phase. But the coding template already parsed it for you. Find the data from our asu google drive shared folder: https://drive.google.com/drive/folders/1W4GLKNsGlgXp7fHtDlhHEBdLVw_IuAXh?usp=sharing. To get the permission to download the data, you have to use your asu gmail account.
-
Zone data (only for hot zone analysis): at "src/resources/zone-hotzone" of the template
The input point data can be any small subset of NYC taxi dataset.
The input point data is a monthly NYC taxi trip dataset (2009-2012) like "yellow_tripdata_2009-01_point.csv"
All zones with their count, sorted by "rectangle" string in an ascending order.
"-73.795658,40.743334,-73.753772,40.779114",1
"-73.797297,40.738291,-73.775740,40.770411",1
"-73.832707,40.620010,-73.746541,40.665414",20
The coordinates of top 50 hotest cells sorted by their G score in a descending order. Note, DO NOT OUTPUT G score.
-7399,4075,15
-7399,4075,29
-7399,4075,22
An example input and answer are put in "testcase" folder of the coding template.
- hotcell-example-answer-withZscore-sample.csv -- answer of the sampled input data with score.
- hotcell-example-answer-withZscore.csv -- answer with score of the full input data from Google Drive shared folder.
- hotcell-example-answer.csv -- the expected output result of using the yellow_tripdata_2009-01_point.csv as the input.
- hotcell-example-input.txt -- the input for running the jar file.
DO NOT DELETE any existing code in the coding template unless you see this "YOU NEED TO CHANGE THIS PART"
In the code template,
- You need to change "HotzoneAnalysis.scala and HotzoneUtils.scala".
- The coding template has loaded the data and wrote the first step, range join query, for you. Please finish the rest of the task.
- The output DataFrame should be sorted by you according to "rectangle" string.
In the code template,
- You need to change "HotcellAnalysis.scala and HotcellUtils.scala".
- The coding template has loaded the data and decided the cell coordinate, x, y, z and their min and max. Please finish the rest of the task.
- The output DataFrame should be sorted by you according to G-score. The coding template will take the first 50 to output. DO NOT OUTPUT G-score.
- Submit your project jar package.
- Zip your project source code and submit to Blackboard.
- Note that: you need to make sure your code can compile and package by entering
sbt clean assembly. We will run the compiled package on our cluster directly using "spark-submit" with parameters. If your code cannot compile and package, you will not receive any points.
This section is same with that in Phase 1.
If you are using the Scala template
- Use IntelliJ Idea with Scala plug-in or any other Scala IDE.
- Replace the logic of User Defined Functions ST_Contains and ST_Within in SpatialQuery.scala.
- Append
.master("local[*]")after.config("spark.some.config.option", "some-value")to tell IDE the master IP is localhost. - In some cases, you may need to go to "build.sbt" file and change
% "provided"to% "compile"in order to debug your code in IDE - Run your code in IDE
- You must revert Step 3 and 4 above and recompile your code before use spark-submit!!!
If you are using the Scala template
- Go to project root folder
- Run
sbt clean assembly. You may need to install sbt in order to run this command. - Find the packaged jar in "./target/scala-2.11/CSE512-Project-Hotspot-Analysis-Template-assembly-0.1.0.jar"
- Submit the jar to Spark using Spark command "./bin/spark-submit". A pseudo code example:
./bin/spark-submit ~/GitHub/CSE512-Project-Hotspot-Analysis-Template/target/scala-2.11/CSE512-Project-Hotspot-Analysis-Template-assembly-0.1.0.jar test/output hotzoneanalysis src/resources/point-hotzone.csv src/resources/zone-hotzone.csv hotcellanalysis src/resources/yellow_tripdata_2009-01_point.csv