Entry Name: " PKU-Lai-MC1"

VAST Challenge 2017
Mini-Challenge 1

Team Members:

Chufan Lai, Peking University, chufan.lai@pku.edu.cn PRIMARY
Qiangqiang Liu, Peking University, qiangqiang.liu@pku.edu.cn
Lu Feng, Peking University, 1601214538@pku.edu.cn
Chenglei Yue, Peking University, chenglyue@gmail.com
Xi Chen, Peking University, chenxi_pku@pku.edu.cn
Yang Hu, Peking University, 1500012738@pku.edu.cn
Zhanyi Wang, Qihoo 360, wangzhanyi@b.360.cn
Pengju Teng, Qihoo 360, tengpengju@b.360.cn
Xiaoru Yuan, Peking University, xiaoru.yuan@pku.edu.cn (Advisor)

Student Team: NO

Tools Used:

EXCEL, MATLAB, and own developed tools based on D3JS.

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

Two weeks, about 300 hours.

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

Video

http://vis.pku.edu.cn/vastchallenge/2017/mc1_video/PKU-Lai-MC1.mp4

Questions

1 - Patterns of Life analyses may also depend on understanding what patterns appear over longer periods of time (in this case, over multiple days). Describe up to six patterns of life that occur over multiple days (including across the entire data set) by vehicles traveling through and within the park. Characterize the patterns by describing the kinds of vehicles participating, their spatial activities (where do they go?), their temporal activities (when does the pattern happen?), and provide a hypothesis of what the pattern represents (for example, many vehicles showing up at the same location each Saturday at the same time may suggest some activity occurring there each Saturday). Please limit your answer to six images and 500 words.

1. To visualize similarity among vehicles, we adopt the dimension reduced projection. Specifically, we transform the trajectories into multi-variate data by taking each sensor as a dimension, and the proportion (in the whole sequence) of a vehicle passing it as the value. For example, if a vehicle passed sensor_1 for 3 times among its 10 sensor-records, we say it has a 0.3 value in dimension 'sensor_1'. In this way, we generate a t-SNE showing the similarity among all 18708 vehicles.

As a result, there are 10 major clusters in the projection, showing 10 different common patterns. Vehicle number of these clusters range from 749 to 847. Vehicles of each cluster took almost the same path.

The 10 patterns share a common feature that all vehicles pass through the park in a short period of time (less than 1 hour), detected by only a small amount of sensors. There are roughly two types of paths in the 10 patterns. In the images below, different colors denote different types of sensors. Beneath each map is a representative sensor sequence of each cluster.

The first type of paths almost directly connect two entrances (except for a few general gates), meaning that the vehicle is probably only passing by the park. Cluster 2('entrance0->general-gate4->general-gate7->entrance1'), 4('entrance0->entrance3'), 7('entrance1->general-gate7->entrance3'), 10('entrance4->entrance2') (as labeled above) belong to such genre.

Another type of paths pass through a few ranger stops except for the entrances and general gates. However, these routes are also passed in a short period of time, meaning that the ranger stops are probably necessary mid-way stops, which we'll further prove in the video. Cluster 1, 3, 5, 6, 8, 9 (as labeld above) belong to such genre. The image below shows the path of cluster 1 as an example.

2. By highlighting the vehicle type in the projection, we find that the 2-axle and 3-axle buses only appear in the large clusters. It suggests that the 10 common patterns may be related to bus routes.

However, except the service trucks typed '2P', all other types of vehicles also participate in these 10 patterns (as well as some other small clusters). It suggests that the 10 corresponding paths are probably major roads going across the park, or popular sightseeing routes taken by sightseeing buses.

3. By highlighting the '2P' types in the projection, we can see that the 998 service vehicles have a similar movement pattern. There are three major types of paths taken by these vehicles, corresponding to three small clusters in the projection. We guest that these vehicles are patrol cars driven by park rangers. From the routes, we can see that different types of paths cover different areas of the park. When combined, they can cover the whole park.

2 - Patterns of Life analyses may also depend on understanding what patterns appear over longer periods of time (in this case, over multiple days). Describe up to six patterns of life that occur over multiple days (including across the entire data set) by vehicles traveling through and within the park. Characterize the patterns by describing the kinds of vehicles participating, their spatial activities (where do they go?), their temporal activities (when does the pattern happen?), and provide a hypothesis of what the pattern represents (for example, many vehicles showing up at the same location each Saturday at the same time may suggest some activity occurring there each Saturday). Please limit your answer to six images and 500 words.

1. A group of 4-axle trucks (containing 23 vehicles) pass through the gates every other time in a fixed route. In fact, they are the only non-2P vehicles that have passed through gates, as described in the video. Since the gates prevent general vehicles that are not Preserve Rangers, we infer that these are 4-axle supply trucks used to carry foods and materials. However, these trucks normally arrived between 2:00 and 6:00 a.m. in the weekdays (Tuesday or Thursday), and stayed for less than 1 hour. Such a strange behavior pattern makes them highly suspicious. If not supply trucks, they may be vehicles quietly transporting and dumping waste materials.

2. The vehicle with ID '20154519024544-322' comes to the park every week for camping (around the camping4 area) during a four-month period from June 19th to October 5th. It comes on Friday and leaves on Monday, staying for 2 to 3 days. It normally enters and exits via entrance4 except for the first time(entrance1 in, entrance4 out). We also find a group of 38 vehicles sharing its movement pattern, but not in a repeated style.

3. The car with ID '20162904122951-717' often comes to the park and camps around the camping0 area, which is within the preserve area. It comes on Friday and leaves on Sunday, staying for 2~3 days. It usually enters and exits via entrance3. This behavior pattern lasts for 6 weeks from March 4th to April 10th. We also find another group sharing this behavior pattern, consisting of 62 vehicles.

4. The car with ID '20154112014114-381' often comes to the park and camps around the camping6 area, which is within the preserve area. It comes on Friday and leaves on Sunday, staying for 2~3 days. It usually enters and exits via entrance0. The pattern lasts for a month from June 12th to July 26th. We also find another group sharing this behavior pattern, consisting of 30 vehicles.

3 - Unusual patterns may be patterns of activity that changes from an established pattern, or are just difficult to explain from what you know of a situation. Describe up to six unusual patterns (either single day or multiple days) and highlight why you find them unusual. Please limit your answer to six images and 500 words.

1. The vehicle with ID '20155705025759-63' is a motorcycle. It entered the park in June 5th 2015, but did not leave until the last record in May 20th, 2016. In other words, it has stayed in the preserve area for almost a year. During its time in the park, it has camped in many different camping areas and stayed in each area for nearly a month. Compared to the suspicious long-term stay, daily movements of this motorcycle are actually quite regular. It was active during the daytime, and quite during the night. Each time it moved, it was moving to another amping area.

2. Some vehicles, such as 20153408043401-757, 20161515101514-592 and 20151818111845-978, have a somehow disordered path. They travel back and forth in some local area without a clear pattern or purporse. These may be self-driving tourists who did not follow popular sightseeing routes, and showed great interests in some specific areas.

3. The vehicle with ID '20163901043959-297' entered the park through entrance0, and leaved after a short visit (about 5 minutes) to general-gate1. We also find a group of 22 vehicles showing a similar patterns with a short visit to camping6.

4. The camping time of some vehicles are longer than a month, such as 20154625114655-96 and 20155517055510-758.

5. We find paths of six cars (including 20152810102819-458, 20152810102803-808, 20152810102828-459, 20152910102928-970, 20152910102959-782 and 20153010103017-871) going through the sequence "entrance1 -> range-stop1 -> range-stop1 -> entrance1". However, we know from the map that range-stop1 is not directly connected to entrance1. It means that these cars probably have cut through the preserve area in their tour, in order not to be deteded by gate2. Also, these cars seem to be a group of friends/famlity that move together.

4 - What are the top 3 patterns you discovered that you suspect could be most impactful to bird life in the nature preserve? (Short text answer)

1. We can see from the projection view that there are many vehicles of various types passing through the park every day. In fact, the 10 major movement patterns all indicate passing-by behaviors, rather than long-time sighseeing or camping behaviors. Given the pollution and noise it causes, such a heavy daily traffic flow could be a great threaten to the habitats of birds.

2. A group of 4-axle trucks pass through the gates every other time in a fixed route, which actually prevents general traffic from passing. They may be illegal to pass, given their active time during the early morning. Hence, these trucks are somehow suspicious.

3. Long-time camping and repeated visits may have influence on the habitats. It's especially suspicious for some people (e.g. 20155705025759-63) to stay in the park for a long time. Park rangers should pay more attention to such visitors to learn about their actual purpose of staying or repeated visiting.