Ranko Miklin,
Tomislav Lipic,
Zoltan
Konyha,
Mario Beric, University of Zagreb,
Wolfgang
Freiler,
Kresimir Matkovic,
Denis Gracanin,
Virgia Tech.,
Student team: YES
For the analysis of data given for this challenge we used the following
tools for noted purposes:
1)
C#
windows application form that we wrote for:
a)
parsing
“Migrant Data.xml” file into csv file
b)
formatting
created csv file to organize data better (extracting year and month from date
and storing them in decimal format, allowing us to define ranges of values – in
this case quarters of the year – we wish to observe as a single entry)
c)
creating
kml file with that shows Encounters in Google Earth system
2)
ComVis
tool – used for analysis of all given data
ComVis was developed at VRVis research center (http://www.vrvis.at). ComVis reads the dataset
from a csv file. Once read and stored user can select one of more views (2D or
3D scatter plot, histogram, pie charts and myriad of others) to represent attributes of the dataset. Besides
scalar attributes as usual in information visualization, ComVis supports function
graphs as attributes as well. All views are linked and simple and composite
brushing is supported. Sessions can be saved to avoid re-reading of the dataset,
to store view set-up, and to easily exchange analysis steps with peers or
create reports for end customers.
3)
Google
Earth – used to display provided map and additional information.
Two Page Summary: YES
VRVis-ComVis-Boats-Summarry.pdf
Video: YES
ANSWERS:
Boat-1 Characterize the choice of landing sites
and their evolution over the three years.
Detailed Answer:
To answer this question we
decided to study following categories:
1)
Record
type (shown in histogram) so we can observe successful landings (confirmed
landing sites) and interdictions (they could give us an idea where the boats
were trying to get) separately.
2)
Encounter
dates shown on few different levels (histograms) – separate entries, months,
quarters and years. We wanted to easily and precisely select large periods of
time but be able to choose finer division if we noticed something interesting.
3)
Encounter
coordinates (2D scatter plot) to see where the encounter occurred
4)
Launch
Coordinates and vessel types but we didn’t find them useful in this challenge.
Neither showed significant deviation from the average (no boat types nor start
locations had much better or poorer interdiction/landing ration).
We learned from Google
Earth (“Island Nations.kmz” and our kml file) that landings were made on shores
of two countries:
1)
USA,
State of Florida, along the shore of peninsula.
2)
Following is location of Isla de Sueno with marked
approximate launch locations.
These screenshots and one
showing encounter coordinates from ComVis,
gave us a very good feeling to which land masses are landing coordinates (red)
related. Interdictions (gray) show where territorial waters of Isla de Sueno
end – parallel in Google Earth.
Isla Mujeras is a popular
tourist destination, but also a jumping point for immigrants trying to reach
Isla Contoy is a guarded
National Park requiring permission to visit. Thus we presume organizers of
migration have contacts in either tour companies or in one of the organizations
running the islad.
Following is the analysis
of landing sites over three years:
First year (2005):
All of the landings were made on the south shore of
Florida – closest one to the Isla de Sueno. Also most of the interdictions took
place in the sea between those two. From this we concluded that this was the
beginning of larger scale migration. Without any experience to draw on, they
chose the most direct route. Most voyages took place between April and August,
whether this was because of weather or some other reason we can’t tell. This
trend continued in following years. These
months, with March and September in lesser degree, experienced the greatest
increase of trips. See Boat- 3 answers
for successful landing rate.
Second year (2006):
There are two important changes in this year in regard to
the last:
1)
West
shore of peninsula is used as a landing site in addition to the south. We
presume that captains decided the routs they used in 2005 are patrolled too
much and tried to find a safer way. This worked in some degree, successful
landing rate for Florida coast increased to 34.11%. Few attempts were made to
reach east coast but all were interdicted, this number is considerably smaller
then of those sailing west.
2)
A large
number of landings (41 out of 129) ware made on coast of Mexico from 15th
of April 2006 onward. Of course waters between Mexico and Isla de Sueno are
rarely (if at all) patrolled by American coastal guard and thus these trips
proved very successful. Considering these landings in addition to those made on
Florida, we get the figure from Boat- 3 answers.
There are many possible reasons for choosing Mexico as a destination among
which we deem following most probable:
a)
Rout to
Mexico is more secure, and from there entry to USA is possible either by land,
or on migrant boats organized by people with more experience.
b)
Deciding
that Mexico is an improvement from Isla de Sueno and Paraiso movement (for
refugees)
c)
Seeking
support and/or resources for Paraiso movement (by supporters)
Third year (2007):
Majority of landings (150 out of 266) were made on
Mexican shore, because of the success it had in 2006. On Florida, south shore
was mostly given up in interest of east and west coasts. Successful landing
rate for Florida increases a little to 36.71%, but Mexico once again improved
the rate considerably. From this we can see that a lot of people decided to
play it safe and avoid the patrols all together. Those who chanced traveling to
USA gave more thought to security then speed favoring the indirect routs.
Boat-2 Characterize the geographical
patterns of interdiction over the three years
Short Answer:
By answering the first
question, we arrived to a better part of this answer. We studied the same data
categories and will reference to parts of the last answer.
First year (2005):
As noted in answer Boat-1
most interdictions took place between Florida and Sueno. These were the usual
patrols expecting immigrants from Cuba and other islands.
Second year (2006):
Many more interdictions are made close to Isla de Sueno –
presumably right outside their territorial waters. This indicates Costal Guard
decided large number of immigrants coming from one place should be sanctioned.
Also a number of
interdictions were made by regular patrols along the west coast – result of route
change noted in Boat-1.
Third year (2007):
Patrols on the route to the west coast have been
strengthened in answer to its interdiction last year. Also there was a number
of arrests on the east since this route was also introduced.
Boat- 3 What is the successful landing rate over
the time period?
Short Answer:
To answer this question, we
used a histogram that shows Record
Type value of dataset entries.
Successful landing rate of all three years is
48.09% - details of histogram show
us the percentage of landings and interdictions in the dataset. See Boat-3-1.jpg (figures are in the
highlighted column).
By introducing another
histogram that shows encounter dates grouped by years and highlighting periods
of time, we got the following information for each separate year:
First year (2005) – 16.58% of records out of which 30.26% are successful landings –
by updating histogram details after the selection of time period we can see
details of records corresponding to the selection (columns S. Br.; S. Br. abs
%; S. Br. rel % in Boat-3-2.jpg) and their proportion with whole date set and
each other.
Second year (2006) – 32.61% of records, 43.14% successful landings
Third year (2007) – 50.82% of records, 57.08% successful landings
