Big Data Analysis

1. Analyzing fishing activity from AIS (Automatic identification system) data

Technologies:

Python
MongoDB

Dataset:

https://zenodo.org/record/1167595#.X_0AltgzaUl
Refers to real data from October 1st,2015 to March 31,2016, in Celtic Sea, Channel and Bay of Biscay (France).

Topics:

Fishing activity report
Fishing vessels trajectory clustering (using Hausdorff distance)
Fishing spots clustering
Finding illegal fishing activities
Clustering algorithms used:
   * KMeans
   * DBSCAN
   * Optics
   * Birch

Files:

FishingAnalysisAIS.pdf (Greek)
nosql_db.ipynb (Add data to local mongodb )
mongodbProject.ipynb 

2. Geospatial Queries on Big Data

Technologies

Apache Spark
Scala

Dataset:

Testing dataset: Hotels and Restaurants Worldwide
Any two geospatial datasets can be used with a few modifications of the code
Data can be downloaded from https://www.dropbox.com/s/dis84tm0r5vyzy7/hotels_restaurants.rar?dl=0

Topics

Geospatial data indexing and partitioning in a cluster
Geospatial join between two datasets given a distance d
  i.e.: Find all hotels close to all restaurants by a distance no more than d
Manual creation of a space-partitioning algorithm (for learning purposes. Well known libraries already exist for this task)
For distance metric, Haversine formula is used

Files

GeoQueries_Spark.jar 
        or
GeoQueries_Spark.scala

Instructions (for jar)

run: spark-submit --master local[*] --class "app"  JAR_FILE d n file1 file2
where:
    JAR_FILE:  jar file location
    d: distance in kilometers
    n: number of partitions to make
    file1,file2: locations of the two datasets to join