Entry Name:? "ISU-Hofmann-MC1"
VAST Challenge
2015
Mini-Challenge 1
Team Members:
Heike Hofmann, Iowa State University, hofmann@iastate.edu?? PRIMARY
Dianne Cook, Iowa State University, dicook@iastate.edu
Eric Hare, Iowa State University, erichare@iastate.edu
Andee
Kaplan, Iowa State University, ajkaplan@iastate.edu
Vianey
Leos-Barajas, Iowa State University, vianey@iastate.edu
Carson Sievert, Iowa State University, sievert@iastate.edu
Samantha Tyner, Iowa State University, sctyner@iastate.edu
Student
Team: NO
Did
you use data from both mini-challenges? We only used the data from MC1 for the analysis below, but we are
using results from the other challenges.
Analytic Tools Used:
R
Approximately how many hours were spent working on
this submission in total?
about 60h.
May we post your submission in the Visual Analytics
Benchmark Repository after VAST Challenge 2015 is complete? YES
Video Download
Video:
https://www.youtube.com/watch?v=z8k8Ursaz7U
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Questions
MC1.1 – Characterize the attendance at DinoFun
World on this weekend. Describe up to twelve different types of groups at the
park on this weekend.?
- How
big is this type of group?
- Where
does this type of group like to go in the park?
- How
common is this type of group?
- What
are your other observations about this type of group?
- What
can you infer about this type of group?
- If
you were to make one improvement to the park to better meet this group?s
needs, what would it be?
Limit your response to
no more than 12 images and 1000 words.
Group
I – everybody in an overview.
Fig.1: Movements
by minute. During the Scott Jones shows the overall number of people on the
move in the park decreases visibly. Right after shows end, there is an increase
in the amount of activity as people leave the area. Quite a few of the rides
end on the hour, leading to the spike in activity right after.
Overall attendance at DinoFun World is characterized
in figure 1. We have plotted the number of movements in each minute (vertical)
along a horizontal time axis and facetted by day. We can learn a couple of
things from this plot:
- the Scott Jones show was held at 10-11 am on all days and on
3-4pm on Friday and Saturday. We can see this from the dip in movements (orange
lines are average number of moves during one hour) for this time, and the
spikes immediately at the end of the show (when a lot of people move out of the
area).
The second show
on Sunday was cancelled, because the dip in movements is missing on Sunday
afternoon. This is also visible in figure 2, showing check-ins by area: people
check into the concert space before 10 am for all days but before 3 pm only on
Friday and Saturday.
- there is a spike in movements on Sunday at around 2:30 pm -
judging from the movement pattern these are people on their way to the Scott
Jones show who get turned away (because there are no check-ins to the concert
space), see figure 3.
Fig.2: Number of
check-ins by minute between Friday and Sunday. The Creighton Pavilion closes
between 9:30 and 11:30 am, and 2:30 and 4:30 pm each day. Check-ins into the
concert space start half an hour before each of the
shows. The second Scott Jones show is cancelled on Sunday, and the Creighton
Pavilion closes at noon on Sunday (presumably shortly after the vandalism is
discovered).
Fig.3: Movement
in the park on Sunday between 2:30 and 3 pm visible as dark dots. Darker and
bigger dots indicate larger amounts of movements. Lots of people try to go to
the Scott Jones show at 3 in the concert space (number 63, South East corner),
but are turned away.
Groups
at the Park
We define a group as a set of people who
rides together. We exclude any of the rides that are potentially walk-throughs
(as determined from the charts showing operating times): Atmosfear,
Galactosaurus Rage and Dykesadactyl
Thrill. Beyond that we include all of the Thrill Rides, Kiddie Rides and Rides
for Everyone.
For each of the days of the time frame we
create a matrix of park goer id and ride id, that consists of 0s and 1s, with a
1 in cell (i,j) indicating
that park goer i was on ride j.
We then use a binary metric to calculate
pairwise distances between park goers as the proportion of rides that only one
of the two individuals took and the total number of rides that at least one of
them did. This yields a distance between 0 and 1, where 0 means that all rides
where done jointly, and 1 that none of the rides was done together.
For example, if individual i took 23 rides, and j did 25 rides, and they did 20 rides
of these together (i.e. in the same 'batch'), the ride based distance between
them is (28 - 20)/28 = 0.2857143.
Using Ward's method for a hierarchical
clustering based on these participant-participant distances, we get groups of
park-goers on Friday, Saturday and Sunday, such as shown at the example of four
groups on Saturday in figure 4. Each individual is shown by a thin grey
horizontal line; for each attraction and ride that a person checks into, a thicker
colored line is drawn from left to right corresponding in length to the time a
person spent in the attraction. When people spend time together in the same attraction,
this results in thick vertical bands. We can see that group 1819 took all rides
together, mainly short ones or ones without a long waiting line, whereas group
1838 did the Flight of the Swingodon twice and the TerrorSaur, all of which had long lines. From the small
white line in the Sauroma Bumpers we see, that one of
the group members sat this ride out. ??
Fig.4: Examples
of four groups on Saturday: the plot summarizes all of the rides that these
folks did together during the course of the day. The two large groups are the
two largest groups on Saturday with 43 and 47 people, respectively.
Fig.5: Summary of
twelve group types.
In
characterizing the type of groups, we use again hierarchical clustering. Figure
5 shows an overview of twelve prominent types of groups.
They can
be characterized as:
- Type 1 are laying low: but they
are taking some kiddie rides.
- Type
2 and 10 are thrill riders, they don?t go into kiddie rides. Type 10 goes
to the Scott Jones show, Type 2 doesn?t.
- Type
3: they don?t go to the Scott Jones show, take a
lot of rides, in particular the Rides for Everyone.
- Type
4: they spend a lot of their time in the thrill rides. Lots of queues for
them.
- Type
5: they came for Scott Jones – and nothing but.
- Type
6: not sure why they came – they don?t do much, but they also don?t
stay long.
- Type
7: Scott Jones!!! They are not as extreme as type 5,
they also do some rides, mostly Kiddie Rides.
- Type
8: Kiddie Rides forever! They also go to some shows in the SabreTooth theatre and enjoy the greenspace.
- Type
9: ?very similar to Type 8, but they
don?t take nearly as many Kiddie Rides.
- Type
11: come late, stay late, go to some shows – they do a little bit of
everything, but they don?t have much time (because they are really coming
in)
- Type
12: they include the huge groups – and they are doing everything
while they are there: Scott Jones show, SabreTooth
theatre, thrill rides, kiddie rides, rides for everyone – everything
but some downtime in the green space, that is.
Improvements!!
Timed
tickets instead of Check-ins?
How
can things be improved for each of the groups? – Most of the groups are
affected by the hour long waiting lines of the thrill
rides and some of the rides for everyone. Figure 6 gives an overview of the
operating times for each one of the thrill rides. On the x-axis we see time of
day, on the y axis the amount of time each individual
spends in the attraction after checking into it. Some attractions seem to be
walk-throughs – such as Atmosfear, Keimosaurus Big Spin, and Wrightiraptor
Mountain; they seem to be fairly short and do not have any(?)
waiting times. All of the other thrill rides, show
massive waiting times at some time during the weekend. The ?drops? of lines
below the solid colored bands indicate individuals who gave up and left the
ride before actually taking it. A different system, such as handing out timed
tickets during rush hours might be less frustrating for all thrill riders.
Fig. 6: Operating times of each of the thrill
rides: on the right time of day is shown, along the y-axis, the amount of time
spent in a ride (defined as the time between check-in and first recorded movement
afterwards). Operating time is shown as a line, each line corresponds to one
individual in a ride. The longer the line, the longer an individual spent in
the ride. This includes waiting times after checking into the ride. It is
obvious, that rides are much busier on Saturday and Sunday. On Saturday or Sunday
park goers have to calculate to take one hour for one ride of the Flight of the
Swingodon, whereas they can manage about 3-4 rides an
hour of the Flight of the Swingodon on Friday. The
pattern of ?droplets? under each of the thick lines indicate people who have
checked into a ride, but leave before it is their turn to take it.
Attention
please, ? General
Announcements?
There
does not seem to be a good system in place to notify park goers about current
events: When the second Scott Jones show is cancelled, hordes of park goers
still try to go to the show (see figure 3), only to get turned away
unsuccessfully. That has to be quite frustrating, and could be easily fixed by
introducing a third park system id (see MC 2.1) that sends announcements to all
park goers. ?
Safety and
Security!
Everybody
is trying to have fun, but some people don?t seem to know when to stop. From
the figures below (figure 7 and 8), it is obvious that some people stay on the
rides and in the attractions longer than expected. The Scholtz
Express – outlier, with a check-in time of 21:53 on Friday and no
movement afterwards, might be a case of a lost communication device (assuming
the best case scenario for the individual), but the three individuals in the
Creighton Pavilion likely spoiled the fun for lots of people on Sunday
afternoon. A few (more) staff members might help with this aspect by gently
shooing folks out of the attractions or checking on them, when their devices don?t
indicate any movements for an unexpectedly long period of time.
MC1.2
– Are
there notable differences in the patterns of activity on in the park across the
three days?? Please describe the notable
difference you see.
Limit your response to no more than 3 images and 300 words.
- The
Scott Jones show is offered twice a day (at 10-11 am and 3-4 pm) on Friday
and Saturday, but only once on Sunday.
Related to that, the Creighton Pavilion is
closed after 12 on Sunday.
- Rides
on Friday are much less busy than over the weekend. Park goers manage to
do more rides on average during the same amount of time.
- On
Friday there are only about half as many people in the park as on the
weekend.
MC1.3
– What
anomalies or unusual patterns do you see? Describe no more than 10 anomalies,
and prioritize those unusual patterns that you think are most likely to be
relevant to the crime.
Limit your response to no more than 10 images and 500 words.
Vandals
in Creighton:
On Sunday, June 8 2014 a trio of suspects (with ids 461004, 416790, and
1502920) checks into Creighton Pavilion at 9:01 am and 9:30 am and stays there
for more than 2 hours, in particular during the time that the Pavilion is
closed to the public. This is likely the time during which the vandalism took
place, and these three are our most likely perpetrators. Figure 7 gives a
comparison of how things usually happen (Friday and Saturday) and what happens
on Sunday.
Fig.7: Vandals
in Creighton Pavilion?! ?Again, on x we have the time of day, and along
y the time each individual spends in the Creighton Pavilion (orange) or at the
Scott Jones show (green). Three individuals stay in Creighton Pavilion during
the time that it is closed to the public.
There are some concert goers who stay extremely long after the show ended or
check in extremely early, but this is likely unrelated to the vandalism.