Entry Name: UBA-Barretto-MC2

VAST Challenge 2015
Mini-Challenge 2

Team Members:

Alfredo Barretto , University of Buenos Aires, acbarrettomdp@gmail.com PRIMARY
Nelson Amaya, University of Buenos Aires, nelmaya@gmail.com

Juan Orlowski, University of Buenos Aires, orlowski@agro.uba.ar

Student Team: YES

Did you use data from both mini-challenges? YES

Analytic Tools Used:

Tableau

Gephi

Excel

R

Sql Server

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

60

May we post your submission in the Visual Analytics Benchmark Repository after VAST Challenge 2015 is complete? YES

Video:

index_archivos\MC2.wmv

Questions

MC2.1 – Identify those IDs that stand out for their large volumes of communication. For each of these IDs

a. Characterize the communication patterns you see.

b. Based on these patterns, what do you hypothesize about these IDs?

There are two IDs that stood out for their large volume of communications (Figure 1). One of them was ID 839736 (Figure 2) and the other one was ID 1278894 (Figure 3).

Figure 1. Communication patterns of the IDs on the weekend

The ID 839736 maintained low and constant levels of outbound-calls during Friday (maximum of 512 calls/hour) and Saturday (maximum of 834 calls/hour) but on Sunday their outbound-calls rocketed reaching a primary peak of 23.654 calls near the time of the crime (12 pm) and a secondary peak of 6.496 calls at 2pm (Figure 2A). The inbound-calls followed nearly the same pattern of Friday (maximum of 511calls/hour) and Saturday (maximum of 832calls/hour). On Sunday, inbound-calls reached maximum values of 23.734 at 12pm and 6.513 at 2pm (Figure 2B) coming from Wet Land and Coaster Alley, respectively (Figure 1C).

Figure 2. Communication patterns of ID 839736 during the studied weekend.

The ID 1278894 maintained regular patterns of outbound-calls in the three days that appeared to be proportional to the attendance of the park. These communications happened at 12pm, 2pm, 4pm, 6pm and 8pm in each day and the amount of calls was around 8.154, 14.760 and 16.908 for Friday, Saturday and Sunday, respectively (Figure 3A). The inbound-calls followed nearly the same pattern again, reaching values of around 8.105 on Friday, 14.684 on Saturday and 16.853 on Sunday (Figure 2B). Although Sunday inbound-calls came mainly from Wet Land, there were also detected inbound-calls from Coaster Alley, Entry Corridor, Kiddie Land and Tundra Land (Figure 2C).

From these patterns we hypothesized that IDs 839736 and 1278894 are robots that register the position of the persons at conflictive periods, mainly, and at regular periods, respectively.

Figure 3. Communication patterns of ID 1278894 during the studied weekend.

MC2.2 – Describe up to 10 communications patterns in the data. Characterize who is communicating, with whom, when and where. If you have more than 10 patterns to report, please prioritize those patterns that are most likely to relate to the crime.

These patterns were found by different techniques. One of the employed techniques was to visualize the communication patterns on Sunday between those persons with an amount of one way communications that represent the observed 5% of the maximum communications between two persons. As a consequence, this technique included patterns between persons that had at least nineteen one way communications between each other. An additional restriction added was that a given person had, at most, communications with other 42 persons. The rationale behind this technique was that if an extraordinary amount of outbound-calls was detected from Wet Land near the time of the crime between few people then the detected pattern should have great chances of being related with the crime because a criminal could have worked alone (in this case no calls should be detected) or with a band of few persons. By this technique five patterns were detected (Figures 3A–E). The persons involved in these patterns made outbound-calls from different areas of the park from early morning (9-10am) to late afternoon (8-10:30pm) but they reached a peak of outbound-calls at around 11:34-11:40am from Wet Land. Some of these patterns involve one and two way communications (Figure 3A,C) or only one way communications, mainly (Figure 3B, D, E).

Figure 4. Communication patterns (A-E) between the most intercommunicated persons on Sunday, who had communications with at most 42 people and who had a peak of outbound-calls from Wet Land at o near the time of the crime. The thick of the arrows represent the amount of communications between the involved persons and their directionalities. Additionally, the color of the dots represent the area of the park where the call was made.

Figure 4 (continued).

Figure 4 (continued).

Figure 4 (continued).

Figure 4 (continued).

The other employed technique was to detect the IDs who produced the most quantity of calls from Wet Land on Sunday, during the period of time when communications rocketed (11:28 am – 12:24 pm). There were detected 37 IDs with the particularity that they intercommunicated only with each other (ID 1635915 called some of them) or, after the mentioned communications, they called to external (IDs 887530, 1635915 and 1708002 are exceptions) and to one or both robots (IDs 668872, 1159870, 1350376 and 1742503 are exceptions) (Figure 4). Some of these IDs have, in general terms, a same pattern of outgoing communications with other IDs that could be used as a fingerprint of the outbound-calls pattern of these given IDs (E.g. IDs 887530 and 955733) (Figure 4).

Figure 5. Communications patterns obtained on Sunday between the IDs with most quantity of outbound-calls from Wet Land during the period of time with most quantity of calls from this zone. Blue, red and yellow patterns are outbound-calls to external, robot ID 839736 and robot ID 1278894, respectively.

Another patterns were founded under a hit and run hypothesis. We search IDs that has low amount off inbound-calls and outbound-calls, and occurs in wet land around the time of the crime (Figures 6 & 7).

Figure 6 Communication pattern of IDs (outbound-calls)

We identified four IDs (Figure 6) that don´t receive any calls and only make one communication to the external (we hypothesize an associated the possibly wait outside the park) from wetland. That is a suspicious behavior that differs from the rest of the IDs. This people were in the place and around the time of the incident. Also, we hypothesize that after the crime the suspects run away.
 

Figure 7 Communication pattern of IDs (inbound and outbound-calls)

Another pattern under a hit and run hipótesis was detected for three IDs that only exchange one call (one inbound and one outbound) with the ID 839736 (the robot system of the park). The calls was made from wetland (place of the crime) and around the time that the vandalism occurs. We hypothesize that after the crime the suspects run away.

MC2.3 – From this data, can you hypothesize when the crime was discovered? Describe your rationale.

We hypothesize that the crime was discovered firstly by visitors and or securities at around 11:28 – 11:39 am because during this period of time the outbound-calls made by people rocketed significantly above the normal amount (Figure 5). Then, between 12 – 1pm robot ID 839736 started tracking the IDs that were at Wet Land (Figure 1C).

Figure 8. Amount of outbound-calls made by persons from Wet Land on Sunday.