MiddlesexMASS-Attfield-MC2

VAST Challenge 2014
Mini-Challenge 2

 

 

Team Members:

Simon Attfield, Middlesex University, s.attfield@mdx.ac.uk  PRIMARY

Peter Passmore, Middlesex University, p.passmore@mdx.ac.uk

Neesha Kodagoda, Middlesex University, n.kodagoda@mdx.ac.uk

Pragya Paudal, Middlesex University, p.paudyal@mdx.ac.uk

David Neilson, Middlesex University, d.neilson@mdx.ac.uk

George Pagiatakis, Middlesex University, g.pagiatakis@mdx.ac.uk

Brian Joyce, Middlesex University, b.joyce@mdx.ac.uk

Robert Wells, Middlesex University, RW489@live.mdx.ac.uk

William Wong, Middlesex Unievrsity, wwong@mdx.ac.uk

 

Adrian Wagstaff, MASS Consultants Ltd., AWagstaff@mass.co.uk

James Bullock, MASS Consultants Ltd, JBulloch@mass.co.uk

Adam Malin, MASS Consultants Ltd, AMalin@mass.co.uk

Dougie Holmes, MASS Consultants Ltd, DHolmes@mass.co.uk

Graham Phillips, MASS Consultants Ltd, GPhillips@mass.co.uk

John Marshall, MASS Consultants Ltd, JMarshall@mass.co.uk

Stewart Bertram, Intelligence Sans Frontieres, berts123@hotmail.co.uk

Student Team:  No

 

Analytic Tools Used:

Patterns of Life Atlas, designed and developed by MASS consultants Ltd and Middlesex University.

Patterns of Life Classifier designed and developed by Middlesex University

Patterns of Life Location TimeLine, designed and developed by Middlesex University.

Excel

SQL

 

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

500

 

May we post your submission in the Visual Analytics Benchmark Repository after VAST Challenge 2014 is complete? Yes


Video

 

 

-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Questions

 

MC2.1


The Patterns of Life (POL) suite was used for analysis. This comprises POL Atlas, which visualises movement and transactions on a map, POL Classifier which is used to label and classify locations and identify trips, and POL Location Timeline, used to visualise where time is spent. 

Figure 1. Using POL Location Timeline this shows time spent for the complete two week period colour coded: GAStech (yellow), home locations (blue), at a coffee bar (light green), at an eateries (brown), at a shop (magenta); other locations (blue). Travel is shown in red. On weekdays, most GAStech employees start the day with a trip to a coffee bar between 6:30 and 8:30, arriving at work between 7:30 and 8:30. They are typically out for lunch between 12:00 to 13:00. They leave work for home between 17:00 and 18:00 and usually go out in the evening to shop or to eat and usually home before 22:00. Weekends are mostly spent at home, although some eat lunch out and perhaps shop and most eat out in the evenings. The colour coded transactions indicate non-driving card users have a similar movement pattern but truck drivers are different. (Red boxes: guide to tool).

Figure 2 Shows the same timeline zoomed in to a typical day, 15th January (but showing all transactions).

Figure 3. Using POL Atlas map shows where trucks go and the transactions made. The view below shows when. The trucks are used only on weekdays. Truck drivers spend the most money (Figure 5) but seldom use credit or loyalty card for personal use. (Red boxes: guide to tool).

 

Figure 4: Morning coffee: this map shows inferred home locations (numbered) and is colour-coded to show who normally goes to which coffee shop, which is local.

.

Figure 5: Summary of credit and loyalty card use.                         

 

MC2.2

Some place locations are inferred. Confidence ratings are given in square brackets. Please see the confidence scale in our response to MC2.3. 

 

1. Alcazar Lucas sometimes works late

On four evenings (6th, 9th, 15th and 18th) Lucas Alcazer’s car (IT Helpdesk 1) leaves GAStech [C3] between 17:30 and 18:00, sometimes goes to his assumed home [C3] or an eatery [B2], but returns to GAStech [C3] after 9pm for 1.5 to 3 hours (see Fig. 1). Assuming he was driving, it seems difficult to explain given his role.

Figure 1. A late night return visit to GAStech [C3] on Mon 6th Jan by Lucas Alcazer’s car, viewed using Patterns of Life Atlas.

 

2. Visit from the Sten Sanjorge Jr, President/CEO

Sten Sanjorge’s (President/CEO 31) car is at Chostus Hotel [B2] from the 6th to 0736h on the 17th when it travels to Hallowed Grounds [B2] (see Fig. 2). We assume prior to that he was out of town with his car kept at the hotel. The visit to Hallowed Grounds [B2] overlaps with Lucas Alcazar’s car (IT Help Desk 1), Lidelse Dedos’ car (Engineering Group Manager 14), and Elsa Orilla’s car (Drill Technician 27). Sten’s car then goes to GAStech [C3], leaving around 1214h for Abila Zacharo [B2]. Also there at that time are Ada Campo-Corrente (SCP/CIO 10), Isia Vann (Perimeter Control 16), Birgitta Frente (Geologist 18) and Adra Nubarron’s car (Geologist 25). They are largely non-regulars at Abila Zacharo [B2]. Sten’s car returns to GAStech [C3] around 1340h. On the 18th, his car visits Katerina’s Café [B2] around 1230h at the same time that Ingrid Barranco (SVP/CEO 4), Inga Ferro (Site Control 13), Varja Lagos  (Badging Office 23) and Anda Riberas’ cars are there. We assume he met these people. The visit is significant as it’s the only time he is in Abila.

 

 

Figure 2. Sten Sanjorge’s car movements on the 17th January, viewed with Patterns of Life Location Timeline. Showing time at Chostus Hotel [B2] (green), GAStech [C3] (yellow) Hallowed Grounds [B2] (orange), Abila Zacharo [B2] (tan) and Katerina’s Café [B2] (brown).

 

3. Executives, including CEO Sten Sanjorge play golf

Ingrid Barranco (SVP/CFO 4), Ada Campo-Corrente (SVP-CIO 10) and Willem Vaso-Pais’ (Environmental Safety Advisor 35) cars go to Desafia Golf Course [B2] on Sunday 12th and again on Sunday the 19th . Sten Sanjorhe (CEO 31) and Orphan Strum’s (SVP/COO 32) cars are there on the19th (see Fig. 3). They each have transactions on their cards of between 110–190 each. We assume they played golf together.

 

After, Ada Campo-Corrente and Willem Vaso-Pais’ cars go to Hippokampos [B2]. Orphan Strum’s car goes there too. We assume they meet for dinner

 

 

Figure 3. Executives’ cars visit Desafio Golf Course [B2] on the 19th January, viewed with Patterns of Life Atlas.

 

4. The CEO Sten Sanjorge Jr goes to Guy’s Gyros

After the visit to Desafia Golf Course [B2] on the 19th, Sanjorge’s (CEO 31) car goes to Guy's Gyros [B2] from 1845h to 1952h (see Fig. 4). We observe overlapping car GPS and credit card transactions for that time with Sven Flecha (IT Technician 17), Stenig Fusil (Building Control 20), Kanon Herrero (Badging Office 22) and Varja Lagos (Badging Office 23). Flecha, Herrero Lagos and Sanjorge pay their bill at the same time and within four minutes of Fusil. Herrero, Lagos and Sanjorge’s cars leave simultaneously. We assume they all had a meeting.

 

Figure 4. Meeting at Guy’s Gyros [B2] on19th.

 

5. Gathering at Lars Azada’s place - Friday the 11th

On Friday 11th there is an apparent gathering at Lars Azada’s (Engineer 2) home [C3] attended by fifteen visitors (see Fig. 5). Most are engineers and IT workers. Their cars begin arriving around 1900h and leaving between 2220h and 0030h. Azada’s car goes to Hippokampos [B2] around 1840h and his card is used for a transaction before returning. Vira Frente’s (Hydraulic Technician 19) car arrives at Azada’s around 1910h and goes to Albert's Fine Clothing [B2] where there is a transaction on her card before returning. We assume this is a party. The gathering is significant as it is the largest outside work.

 

 

Figure 5: Azada’s gathering, viewed with Patterns of Life Locations Timeline. Showing time at Azada’s home [C3] (green), Albert’s Fine Clothing [B2] (light brown).  

 

6. Isande Borrasca and Brand Tempestad apparently meet for regular lunches at the Chostus Hotel

Isande Borrasca (Drill technician 7) and Brand Tempestad’s (Drill Technician 33) cars go to the Chostus Hotel [B2] on the 8th, 10th, 14th and 17th (see Fig. 6). They arrive between 1100h - 1130h and stay for two to three hours. They spend over 100 each every time. We assume they have lunch. This may be significant because they use separate cars even though their routes are the same, suggesting something clandestine.

 

 

Figure 6: Isande Borrasca and Brand Tempestads’ visits to the Chostus Hotel, viewed with Patterns of Life Locations Timeline. Showing time at hotel (green) and time at GAStech for reference (yellow).   

 

7. Hennie Osvaldo night-time behaviour

After work, Hennie Osvaldo’s (Perimiter Control 21) car goes to Lidelse Dedos (Engineering Group Manager 14) and Birgitta Frente’s (Geologist 18) assumed home [C3].  He stays over on the two Wednesdays and the weekends. Each Tuesday, Thursday and Friday and on Monday 6th, Hennie’s car goes to where Inga Ferro (13), Loreto Bodrogi (15) and Isia Vann (16 - all security) apparently live [C3] (see Figure 7).

 

Figure 7. Showing Hennie Osvaldo’s visits (line 21) to Lidelse Dedos and Birgitta Frente [C3] (blue), Loreto Bodroqi and Isia Vann [C3] (red)

 

8. Night visits to executives’ houses

Security guard cars make early hours visits to executive’s assumed homes on four occasions (see Fig. 8). The visits are: 7th - Ada Campo-Corrente’s home [C3] (SVP/ICO 10); 9th - Strum Orhan’s home [C3] (SVP/COO 32); 11th - Willem Vasco-Pais home [C3] (Environmental Safety Advisor 35); and 14th - Ingrid Barranco (SVP/CFO 4) [C3]. Each is visited once. This is significant because all these people work during the day (i.e. it is probably not night shift work). 

   

 

 

 

Figure 8: Night visits by security guards to Ada Campo-Corrente’s [C3] (green), Strum Orhan’s [C3] (orange), Willem Vasco-Pais’ [C3] and Ingrid Barranco’s [C3] (blue), viewed with Patterns of Life Locations Timeline.

 

 

9. Trucks behaving strangely

From the 16th, the trucks begin some strange behaviours (see Fig. 9). They drive in circles or back and forth continuously, repeating routes without stopping along parts of their normal routes. The trips last between 30 minutes and four and a half hours. Four trucks begin this simultaneously at 1919h on the 16th from different locations. They are likely driven by Benito Hawelon, Henk Mies, Valeria Morlun, and Dylan Scozzese. The return to GAStech [C3] is staggered. After a short break Henk Mies goes out for a second trip. On 17th, a truck probably driven by Cecilia Morluniau and possibly with Irene Nant does similar, driving in loops. This is significant because it is not a normal pattern.

 

Figure 9. Truck movements on 16th, viewed using Patterns of Life Atlas. Repeating colour patterns on the time line show repeating journey patterns from 1919hrs. 

 

10 Loyalty Card/Credit Card irregularities

Around 15% of credit card transactions (219/1394) are mismatched with a loyalty card transactions the same day, where the loyalty card amount is either exactly 20, 40, 60, or 80 units less than the credit card amount. This affects 51 out of 54 known people, occurs at most shops (marked in red on Fig 10), and occurs throughout the period. This is significant because it is systematic error may indicate a fraud.

A further significant loyalty card anomaly is the ~23000 claimed on loyalty cards with no matching credit transaction suggesting cash payments not for business by six truck drivers: Scozzese, Benito, Hawelon, Mies, Nant, Valeria Morlun, and Morluniau. A less significant anomaly is the use of Elsa Orilla’s (Drill Technician 28) loyalty card for fifteen small payments by Kanon Herrero (Badging Office 22). 

  .

 

Figure 10.  Map showing locations in red where there are loyalty card anomalies, viewed using Patterns of Life Classifier. 

 

11 Security Visits to Unidentified Locations C1-C6.

 

Six unknown locations (C1-6 on Fig. 10) are visited by security personnel only. Mostly, these visits occur around noon (see Fig. 11).

 

 

Figure 11. Security guard visits to C1-6, viewed using Patterns of Life Location Timeline. 

 

12. Ovan Bertrand’s car visits C3 at night

Ovan Bertrand (Facilities Manager 29) visits one of the unknown locations (C3) at night – around 2325h on Sat 11th. He spends just under 30mins there, then head to his assumed home [C3] (see Fig 12). This is a location only visited by security guards. He is the only one to go there at night. It comes at the end of a busy two hours in which he visits Guy’s Gyros [B2], Ouzeri Eilan [B2], Kalami Kafenion [B2], and Hippocampus [B2].

 

 

Figure 12. Ovan Bertrand’s car visits C3 at night for around 30 minutes, viewed using Patterns of Life Atlas.

 

 

MC2.3

Problematic data was identified by plotting.

 Figure 1. Plot of 28’s GPS data.  This was corrected by averaging and then translating the data to align with the GAStech location.

Figure 2. Non-contiguous journeys. Axel Calzas (9) was found to make journeys that did not start and end in the same place.  Between journeys had to be inferred.

Locations were derived by either:

(a) Using credit card and GPS data and using the median of coordinates at transaction time; (b) Inferring from GPS and possibily map data or inferring by the user or by credit card use; or (C) Inferred from map data (by colocation of the tourist data with the Abila street map).

Unknown locations visited by trucks were resolved through their credit card use. Identification of which driver drove which truck was then inferred from location and credit card use.

Example inference, Brewed Awakenings.

Figure 3. Plot of GPS positions at time of credit card transactions for Brewed Awakenings can’t be used for location.

Resolved: Ingrid Barranco (4) stopped at Brewed Awakenings on the way to work. The credit card timestamp was invalid but the location could still be found using the ATLAS timeline and map.

Figure 4. Gap in timeline implies location.

The following confidence scale was used for locations.

Source evaluation

Intelligence evaluation

A

Always reliable

1

Assumed true as provided by VAST

B

Mostly reliable

2

Assumed true but may contain errors

C

Sometimes reliable

3

Inferred by analysis team so may contain errors

D

Unreliable

4

Cannot be judged

E

Untested source

5

Suspected to be erroneous

 

Figure 5. With inclusion of 6 new labels (C1-6) all trips and locations were classified. Locations derived from the tourist map were classified B4 (green), VAST supplied data B2 (red), and inferred data C3 (black).