Entry Name: "SMU-Kam-MC2"

VAST Challenge 2017
Mini-Challenge 2

Team Members:

Hongjun Qian, Singapore Management University, hongjun.qian.2016@mitb.smu.edu.sg
Jiaqi Zhang, Singapore Management University, jiaqi.zhang.2016@mitb.smu.edu.sg PRIMARY

Xintian Liu, Singapore Management University, xintian.liu.2016@mitb.smu.edu.sg

Student Team: YES

Tools Used:

Tableau

Excel

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

50

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

Video

Yes (In the file folder)

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.

1) Improperly Working

· Stop Working

The graph above shows an overview of all the chemical readings of 9 monitors in the three months. The color represents the amount of the chemicals that monitors detected. However, we can also find that there are some blank grids which means there are no readings of chemicals at 0:00 am. This means the monitors are not working properly at 0:00 am these days. The table below shows the missing value of each sensor, which means they are not working at 0:00am on these days. The reasons may be that the sensors are in maintenance period or some other conditions that cannot work. (work = “Y”, Not work = “N”)

· Wrongly Record

The graph above is an example showing the readings of monitor5 in April6, April 7 and April8. We can see from the graph there are some gaps in the readings of Methylosmolene at some periods. However, each time when there are gaps in Methylosmolene, the readings of AGOC-3A always increased dramatically. Actually, these are gaps are very common also in other monitors and in other dates. This strange pattern inspires us to detect further for the reason of this phenomenon.

The bar chart above shows the count of monitor 5 readings in any specific hours. If we lock it into entire view, we can easily find that the range of “CNT(reading)” is 0-2. What’s more, the condition of “CNT(reading)=2” only appears in one chemical---Appluimonia and “CNT(reading)=0” only appears in the other chemical---Methylosmolene. The overlap and the gap are exactly corresponding, which means, the reason for the strange pattern mentioned above is because monitors wrongly recorded some value of “Methylosmolene” to “Appluimonia”.

The table below shows the information of wrong records of different sensor in three months. (Recorded wrongly= “×”, recorded correctly = “√”).

2) Unexpected Behaviors

· Some Extreme High Readings

As can be seen in the graph above which is the readings of monitor1 in the 3 months, some readings are extreme high in the range of its corresponding monitors. The following table summarizes some extreme high readings for different monitors. These high readings are also questionable because they are quite outstanding from the normal range. The reason may be the sensors were not working properly or there were some external environmental factors such as wind. (The high value of AGOC-3A can be ignored here because these are because the monitor recorded AGOC-3A twice)

· Some Other High Readings

Except for some extreme high readings, there are also some other high readings in the normal range. For example, the graph above shows the readings of monitor 1 in the middle of April. We can see that for each chemical there’s a reference line showing the median with quartiles and the median value of Chlorodinine is 0.199 but the value in April 16 11:00 is at 5.090. This is a high value compared to the median value and needs further investigation. Especially, we will put more focus on the reading of chemical Chlorodinine because Corrosives are materials that can attack and chemically destroy exposed body tissues. It is a dangerous chemical to our environment and human beings. The other thing need to mention is that monitor 3 always fluctuated in the readings and it showed the most variation.

Some other high readings are shown in the following table.

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.

1) Sensor vs Chemicals

The graph above shows an overall pattern of the reading of different chemicals in 9 monitors in the three months. As can be seen, monitor3 always show high readings in the four chemicals in the three months. The readings on monitor 4 are small in April but increased dramatically to be the highest in all the chemical readings in December. Monitor 6 shows a high reading of Methylosmolene and AGOC-3A in April while monitor5 and monitor6 has high reading of AGOC-3A in August.

2) Release Patterns

For the release pattern, we start from detecting if there are any trends in different months. Then we’ll break down to release pattern on different dates and hours to see how these factors affect the release readings.

The graph above shows the chemical release pattern in different months. The color represents readings of the monitors. The stronger the color, the higher amount of the chemical was released. Again, it is very clear that for monitor4, the release for all the four chemicals increase from April to December. Monitor 3 always detect high release in the three months but the release of AGOC-3A in August is higher than other months. Monitor 6 shows a more casual pattern of the release of different chemicals as well as in the release time.

In terms of the release date of the chemicals, we can see from this calendar graph that some monitors have high readings on several dates. For example, monitor 3 shows a high release on the second Saturday in April and on the first two Mondays in December. Monitor 4 detected that the release was high in first Monday and third Sunday in December. On the second Friday in August, the release was quite high showing in monitor 5. For monitor 6, again it shows some casual pattern of release because the dates with high readings are always changing among three months.

Then we break down more to focus on the hourly release pattern of different monitors. We can see that the release of AGOC-3A was always high from 5:00 am to 14:00 pm in April while the release of Methylosmolene was high mainly in the early time in one day staring from 0:00 to 6:00. The release of AGOC-3A shows the same pattern in August and December that it released from 6:00 am to almost 20:00 at night but the release amount in April is much higher than in December. The release of Methylosmolene was high from 0:00 to 6:00 in December.

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.

We need a visualized graph to show the wind direction, wind speed and the released chemicals by each factory. The graph below is a visualized graph of wind information and conditions of chemical release at every specific time. The radius of the fan-shape is related to wind speed and the minutes the wind blows. The overlap means the areas are polluted by the chemicals. The pie-chart on each sensor means the proportion of different chemicals that each sensor detects. The minutes that wind blows and the degree of the wind range can be filtered in this application. For example, this is 0:00am on Apr 1st, if the wind angle range is 60 degrees and if the wind blows for 9 minutes, monitor 9 can detect the pollutants from Radiance Colour Tek and Indigo Sol Boards, while monitor 4 can detect the pollutants from Roaddrunner but not Kasios Office Furniture.

1) Basic Concepts and Assumptions

At first, there are some concepts of wind need to be mentioned.

• We have the wind speed and wind direction every 3 hours, but the sensor reading updates every hour. So, we assume that the meteorological data at 3*N o’clock will influence the sensor readings at 3*N-1, 3*N and 3*N+1 o’clock.

• The wind can both carry and blow away the pollutants, which is determined by the wind speed and minutes. For example, in this graph, monitor 4, 5 and 9 are in the overlap of wind. As normal, the readings at 6:00am, Apr 1st should be high, but in fact, they are not. This is caused by fast wind speed or factories not releasing.

2) Chemicals vs Factories

“AGOC-3A” is released by “Radiance Colour Tex”. We can find that at 3:00, Apr 11th, monitor 9 has a high value of “AGOC-3A” and monitor 5 has a low value of it. Around one hour later, monitor 3, 4, 5 and 9 is overlapped, but the reading of monitor 5 increases while monitor 9 decreases which means “AGOC-3A” is not moving to monitor 9 but 5. Also, we can find that in the pie chart of sensor 9, “AGOC-3A” and “Methylosmolene” are both detected. So, “AGOC-3A” and “Methylosmolene” are released by “Radiance Colour Tex”.

“Appluimonia” is released by “Indigo Sol Boards”. We can find in this specific time, Sensor 5 is overlapped by the pollutants released by “Radiance ColourTek” and “Indigo Sol Boards”, while sensor 9 is only related to the pollutants released by “Indigo Sol Boards”. From the readings of “Applumonia” in sensor 5 and 9, we easily find that “Appluimonia” is mostly released by “Indigo Sol Boards”.

“Chlorodinine” is mostly released by “Roadrunner Fitness Electronics”. In the graph above, Sensor 4 is overlapped by the pollutants released by “Roadrunner Fitness Electronics” and “Kasios Office Furniture”, while sensor 3 is only related to the pollutants released by “Kasios Office Furniture”. From the readings of “Chlorodinine” in these two sensors, we find that “Chlorodine” is mostly released by “Roadrunner Fitness Electronics”.

“Methylosmolene” is released by “Kasio Office Furniture”. In this graph, sensor 6 can detect the pollutants released by “Roadrunner Fitness Electronics” and “Kasios Office Furniture, while sensor 9 can only detect Roadrunner Fitness Electronics’ pollutants. The “Methylosmolene” reading of sensor 6 is much more than that of sensor 9. So, “Methylosmolene” is mostly released by “Kasio Office Furniture”.

After we check the wind visualization graph and conclude all the patterns between factories and chemicals, the results are shown below.