This form is a web page which was created in MS WORD and therefore can be easily edited that way

Entry Name: UIC-Marai-MC2

VAST Challenge 2017
Mini-Challenge 2

Team Members:

Joshua Castor, University of Illinois at Chicago, jcasto3@uic.edu PRIMARY

Joseph Borowicz, University of Illinois at Chicago, jborow5@uic.edu

Andrew Burks, University of Illinois at Chicago, aburks3@uic.edu

Manu Thomas, University of Illinois at Chicago, mthoma52@uic.edu

Timothy Luciani, University of Illinois at Chicago, tlucia2@uic.edu

G.E. Marai, University of Illinois at Chicago, g.elisabeta.marai@gmail.com

Student Team: YES

Tools Used:

Excel

d3.js

Bootstrap

Google Chrome

SAGE2

VAST Challenge 2017 - Mini Challenge 2 was developed by undergraduate researchers (REUs) at the Electronic Visualization Lab, University of Illinois at Chicago

Approximately how many hours were spent working on this submission in total?

100+ hours

May we post your submission in the Visual Analytics Benchmark Repository after VAST Challenge 2017 is complete? YES

Video

· Public Link

Images

· Public Link

Questions

MC2.1 – Characterize the sensors’ performance and operation. Are they all working properly at all times? Can you detect any unexpected behaviors of the sensors through analyzing the readings they capture? Limit your response to no more than 9 images and 1000 words.

A. Our solution is a web-based visual analysis tool, with linked views that show:

A map with the physical location of factories (circles) and sensors (square glyphs), and optional streamlines that reconstruct particle pathways over time. A constant diffusion factor is added to the resulting pathways to show areas of possible emissions.
An overview, calendar timeline at the top
A detailed chemical chart view on the right, with the selected sensor data shown in color, and the other sensor data in gray
Additional filters that control the streamline simulation and the information shown.

Colors are mapped to the 4 different chemicals. Brushing and linking through color and interaction are used to correlate the views.

The sensors are not performing properly at all times. For example, we found a time period from Aug 1st 2016 0:00 to Aug 4th 2016 18:00 where we had no wind data (skipped data, not 0), indicating that the wind sensor wasn’t functioning properly in that period. The lack of wind data is automatically detected in our tool, and clearly indicated.

B. The sensors are grayed out (or partially grayed out) and marked with a red cross when we detect missing chemical readers from those sensors. There are multiple instances, including the 2nd and 7th of every given month, where we were missing chemical readings (partially or completely) for nearly all of the sensors.

C. During the period of no wind data, sharp spikes were observed in Appluimonia and Chlorodinine in sensors 2 and 6.

D. The Appluimonia and Chlorodinine readings for sensor 3, and to a lesser extent sensor 7, seem to fluctuate a lot more compared to the other sensors. This could indicate malfunctioning sensors, as the sensors around sensors 2 and 4 don’t fluctuate as much.

E. In multiple cases, when Methylosmolene has no reading, AGOC-3A has 2 readings. For example, at timestamp 8/8/16 10:00, Methylosmolene has a null reading, which indicates that there’s no entry for it, while AGOC-3A has two readings. What’s interesting to note is that sometimes there will be a large difference in the two values for AGOC-3A, as shown by the 69.17 ppm reading and 7.08 ppm reading for AGOC-3A on 4/6/16 6:00.

MC2.2 – Now turn your attention to the chemicals themselves. Which chemicals are being detected by the sensor group? What patterns of chemical releases do you see, as being reported in the data?

Limit your response to no more than 6 images and 500 words.

a. Appluimonia was detected by all the sensors, though sensors 1 and 2 didn’t detect as many peaks compared to the rest.

b. Chlorodinine concentration spikes were detected by all sensors except sensor 9.

c. Methylosmolene concentration spikes were detected by all the sensors except 1.

d. AGOC-3A was detected by sensors 5, 6, and a little bit by 9.

e. Starting in August and into December, there seems to be a slowly rising concentration for Appluimonia and Chlorodinine for Sensor 4

f. August seems to be the month where we have the least amount of peaks for Methylosmolene and AGOC-3A.

g. For all three months, a majority of the peaks in chemical readings for Methylosmolene happen in the midnight hours (sometime between 22:00 and 5:00)

MC2.3 – Which factories are responsible for which chemical releases? Carefully describe how you determined this using all the data you have available. For the factories you identified, describe any observed patterns of operation revealed in the data.

Limit your response to no more than 8 images and 1000 words.

The streamline simulation (that we created to analyze this data) starts with one seed point at each factory. For each timestamp simulated, a wind vector is added to the calculations, transposing the previous points to create a line. We include four wind vector options, because the wind data is only available for nearly every third hour. The options include using the last available wind data point, the next available data point, the closest available data point, and an interpolation between closest available data points. We add wind direction to the magnitude to create the offset vectors. A constant diffusion factor is added to the resulting path to show an area of possible emissions.

Roadrunner Fitness Electronics (RFE) is the primary polluter of Chlorodinine. This is shown by a streamline simulation using wind interpolation starting from 12/22/16 16:00. This simulation shows peaks in sensor 6 at 12/23/16 0:00 and 12/23/16 5:00, and in both these cases the streamline from RFE intersected sensor 6. Similarly, we have a streamline simulations starting from 8/11/16 10:00 and stopping the time slider on 8/11/16 22:00, 8/12/16 2:00, and 8/15/16 10:00 that capture similar behavior.

Methylosmolene’s primary polluter is Kasios Office Furniture (KOF). On, 4/2 4:00 the streamline emanating from KOF intersects sensor 6 showing a concentration in excess of 88 ppm. At 4/3 0:00 the streamline intersects sensor 6 showing a concentration of 42 ppm. Similarly, the streamline simulation at 4/9 1:00 clearly intersects the sensor 6 showing a concentration in excess of 94 parts per million of Methylosmolene.

Radiance Colourtek (RC) is the primary polluter of the chemical AGOC-3A. Beginning 4/15/16 6:00 and extending to 4/15/16 12:00 the streamlines emanating from RC that intersect sensor 6 show spike AGOC-3A concentrations that exceed 45 ppm. The AGOC-3A peaks on August 6th, 12th, and the 25th all have intersecting streamlines emanating from RC as well. Finally, the largest spike in concentration of AGOC-3A in December occurred on the 15th at sensor 6 @4:00 also had a streamline starting at RC intersecting at that point.

RC’s secondary pollutant is Applumonia, shown by streamline simulations through 12/7/16 1:00, 12/26/16 10:00, and 8/13/16.

We would be wary of anything using sensor 3 as a piece of evidence, as we noted earlier that it could be a malfunctioning sensor.

Indigo Sol Boards (ISB) primary pollutant is Appluimonia. On 4/7 1:00 and 2:00, sensor 9 shows peaks of Appluimonia in excess of 4.9 ppm clearly has an interesting streamline originating from ISB. This situation is repeated on August 10th. Similarly a streamline originating at ISB on 12/15 intersecting sensor 9 during concentration peaks at 12:00 and 21:00.