IS428 AY2019-20T1 Assign Fresi

Problem & Motivation

St. Himark is a vibrant community with a population of 246,839 people located in the Oceanus Sea. It also home to the Always Safe nuclear power plant, a 200 Megawatts electric (MWe) pressurized water reactor (PWR). The Always Safe nuclear power plant provides 72% of the city’s electricity and provides jobs for 700 highly skilled professionals. It has always been the pride of the city, yet it was not quite built to be compliant with the recommended safety standards developed by international organizations and is now ageing.

The nuclear plant sustained moderate (but not life-threatening) damage during the earthquake resulting in some contamination being released. Further, a coolant leak sprayed employees’ cars and contaminated them at varying levels. The city has installed stationary sensors which could be used to determine affected areas, but they don’t cover all parts of the city. Luckily, the Himark Science Society has stepped in with citizen scientists to help crowdsource this important task. Utilize these various sensors to understand which parts of the city are most affected. Meanwhile, mayor Jordan, city officials, and emergency services are overwhelmed and are desperate for assistance in understanding the true situation on the ground and how best to deploy the limited resources available to this relatively small community.

With the huge amount of contamination data collected from the sensors reading, there is a need to build an interactive comprehensive visualization platform to monitor which parts of the city are the most affected.

The interactive visualization will focus on:

• The monitoring of Static sensors reading, one at the entrance to the Always Safe nuclear power plant and eight dispersed around the city.
• The monitoring of 50 mobile sensors that travel in and out of the city.
• Understanding the observation and identify the anomalies from the static and mobile sensor reading.
• Identifying the area/s that are heavily affected by the contamination and dispatch necessary help for its citizens.

Dataset Analysis & Transformation Process

The data zip folder contains 1 shapefile folder, 4 maps of St Himark city, and 3 CSV files.

The 3 CSV files are:

• StaticSensorLocations.csv
• StaticSensorReadings.csv
• MobileSensorReadings.csv

Static Sensors

There are 9 static sensors in St. Himark based on the picture above, where one sensor (sensor-id 15) is placed at the entrance to the Always Safe nuclear power plant and eight dispersed around the city. These detectors provide near-real-time monitoring in the unlikely event of a radiological emergency.

Looking on the static sensor location data, we notice a problem.

Issue: The StaticSensorReadings lacks of location-id and the neighbourhood attributes

Solution: Using the StaticSensorLocations and Figure 1, we update the record by adding additional 2 column which are (1) location-id and (2) neighbourhood.

Mobile Sensors

There are 50 mobile sensors attached to the vehicles. The timestamps are reported in 5-second intervals. The locations travelled by the mobile sensors are based on the longitude and latitude values.

Issue: Like StaticSensorLocations.csv’s problem, the MobileSensorReadings.csv only has Longitude and Latitude attributes to match the vehicle with its location.

Solution: Combine the MobileSensorReading.csv to St.Himark.sph by using inner join function in Tableau.

We use inner join (Intersects) between Join calculation of MAKEPOINT([Lat], [Long]) on MobileSensorReadings.csv and Geometry on St.Himark.shp.

Once done, Tableau will produce three additional attributes, which are (1) Id – neighbourhood id number, (2) Neighbourhood, and (3) Geometry.

Combining Static and Mobile Data

Next, we need to combine the static and the mobile sensor data together. The structure of the captured data by both the static sensor and mobile sensor is similar. However, the static sensor doesn’t have a longitude and latitude attributes and therefore unable to identify the exact location of the sensor. Below image shows the main differences of attributes between static and mobile data captured:

The following section illustrates the issues faced in the data analysis phase and the data transformation needed to create the desired data structure used for insight generation.

Issue: The details of Longitude and Latitude of the static sensor is presented in the StaticSensorLocation instead of StaticSensorReading. We need to combine the two data together to enable the visualization of static sensor location on the St. Himark’s map.

Solution: Combine the StaticSensorLocation file with the StaticSensorReading file by using inner joint of sensor-id in Tableau Prep.

Once the data is joined, there will be a duplicate of Sensor-id1 attributes which we removed afterwards through the cleaning process. At the end in the Output 2 (picture below), we have the StaticSensorReading with the Latitude and Longitude attributes.

Before we combine Mobile and Static sensors data, we need to rename the id column and Nbrhood attribute in the mobile sensors file (name: Extract (Extract)). We rename id to Location-id and Nbrhood to Neighbourhood for easy understanding of the data.

Issue: Once, the union between static and mobile sensor is formed, it will be hard to distinguish if a sensor is a static or a mobile sensor especially the number of static sensor-id is the same as mobile sensor-id.

Solution: Rename the Table Name attribute from the union table to “Sensor Type” while we change the 2 values from MobileSensorReading.csv and StaticSensorReading.csv to “Static” and “Mobile” respectively. Afterwards, we removed the Units attribute as both static and mobile has the same measurement of CpM. We also trim spaces in User-id attribute as some input has extra space.

Lastly, we will extract the finalized data that contains both static and mobile sensor reading.

Interactive Visualization

The visualization platform can be accessed here: TO BE UPDATED

Interesting & Anomalies Observations

Using the graphs from the visualization platform to perform investigation and analysis, the following aims to provide the answer to the questions posed.

Q1: Visualize radiation measurements over time from both static and mobile sensors to identify areas where radiation over background is detected. Characterize changes over time. Limit your response to 6 images and 500 words. Provide your answer and corresponding images here.

Sensor Type	Radiation Measures	Proof
Static Sensor	There is an increase of radiation measures recorded from 8 April 2020 at approximately 1500. The radiation level is persisted until the end of the recording period.	Figure 1.1
Static Sensor	Based on Figure 1.2, all the static sensors displayed constant radiation with minor random variation within the healthy range of 75 CpM and random but infrequent spikes from 6 April 2020 at 1200 to around 8 April 2020 at 1600. From 8 April 2020 at approximately 16:00, sensor-id 15 which is the nearest to the nuclear plant showed a significant increase in radiation level. It followed by the raised of radiation in sensor-id 12, 13, 14 (all 3 are located near to the power plant) and 15 for around 20 minutes. From 9 April 2020 at approximately 1430 to the end of the recording period, there were elevated background radiation from all sensor except for sensor-id 15.	Figure 1.2
Static Sensor	Based on Figure 1.3, again, we spot the slight rise of radiation level in sensor-id 15 on 8 April 2020 at approximately 1600, followed by the rise of radiation level in sensor-id 12 (Jade Bridge) at approximately 1700 on the same day. Sensor-id 9 (Old Town) and 11 (Broadview) displayed an increase of radiation level on 8 April 2020 from 2200 to 9 April 2020 at 0600.	Figure 1.3
Mobile Sensor	We spot obvious patterns in the mobile radiation reading as followed: Sensor-id Area Pattern Timestamp 7, 8, 10, 11, 12, 46 Jade Bridge and Old Town Increase in radiation level 9 April 2020 from approximately 0700 - 1930 7, 8, 10, 11, 12, 46 Jade Bridge and Old Town Decrease in radiation level 10 April 2020 from 0700 - 2130 5, 9, 14 Jade Bridge High and stable radiation level 9 April 2020 at 0800 to 10 April 2020 at 2030 21, 22, 24, 25, 27, 28, 29, 30, 45 Wilson Forest Highway Increase in radiation level 9 April 2020 at 1930 to 10 April 2020 at 0830	Figure 1.4
Mobile Sensor	Mobile sensor-id 10 (Jade Bridge) showed the highest amount of radiation level after the earthquake hit St. Himark. The rise of radiation began from 8 April 2020 at about 1700 to 2100. As the contamintation is spread, mobile sensor-ids (5, 7, 8, 9, 10, 11, 12, 14, and 46) recorded a high radiation when passing through the Jade Bridge area as what we could observed from Figure 1.3. From 9 April 2020 at approximately 1900 to 10 April 2020 at 0600, mobile sensor-id 21, 22, 24, 25, 27, 28, 29, and 45 which pass through the Wilson Forest Highway recorded a substantial increase in radiation level. The increment in these sensors is aligned with the observation from Figure 1.3.	Figure 1.5
Mobile Sensor	Sensor-id 9 showed an increase of radiation level for about 2 hours on 8 April 2020. It followed by the rise of radiation level on sensor-id 11 which lasted for 5 hours on the same day. On 9 April 2020 from 1800 onwards, sensor-id 29, 45, 21, 25, 27, 28, 22 and 24 recorded a huge increase of radiation level.	Figure 1.6

Q2 – Use visual analytics to represent and analyse uncertainty in the measurement of radiation across the city.

a. Compare uncertainty of the static sensors to the mobile sensors. What anomalies can you see? Are there sensors that are too uncertain to trust?

Sensor Type	Uncertainties and Anomalies	Evidence
Static Sensor	Referring to Figure 1.3, sensors-id 15 and 12 recorded an increase in radiation starting from 8 April 2020 at around 1600. Subsequently, all static sensors displayed a surge of radiation except for sensor-id 15 which stopped recording from 8 April 2020 at 2200 to 10 April 2020 at 2000 (based on Figure 1.2). Looking at the distribution of static sensors boxplot, we see static sensor-id 12 displayed an extreme radiation on 8 April 2020. By 10 April 2020, there are 2 categories of outlier, (1) sensors-id 9 and 12 showed highest value of radiation and (2) sensor-id 15 recorded an extreme low value given the proximity of the sensor to the nuclear power plant. Except for sensor-id 15 which implied low certainty due to the absent of radiation reading from 8 April 2020 at 2200 to 10 April 2020 at 2000, all static sensors-id offer a consistent and reliable radiation measures over time given the condition of the city.	Figure 2.1
Mobile Sensor	Mobile sensors present a more uncertainty in the radiation measures over time. The first anomaly is there are 10 mobile sensors-id (1, 6, 23, 26, 34, 35, 47, 48 and 49) which had stopped recording on 8 April 2020 at around 0719 onwards as what shown in Figure 2.2.	Figure 2.2
Mobile Sensor	Third, mobile sensors 18, 20, 38, 39, 43, 44, 47 and 48 displayed random variation and spikes of varied radiation measures, which should be within the healthy radiation bound before the quake happened. Also, sensor-id 18 stopped recording from 8 April 2020, 1623 to 9 April 1243. Last, sensor-id 43 showed a discrete drop in radiation from 10 April 1320 to 1535. /center>	Figure 2.3
Mobile Sensor	Sensors-id 12, 19, 20 and 21 displayed a substantial background noise (low persistent spike below 75CpM) of radiation measure. In conclusion, mobile sensors-id provide a less consistent and therefore less reliable radiation measure. More considerations need to be taken to draw conclusion based on the mentioned mobile sensors-id mentioned above.	Figure 2.4

b. Which regions of the city have greater uncertainty of radiation measurement? Use visual analytics to explain your rationale.

Sensor Type	Regions with uncertainties	Evidence
Static Sensor	For the static sensors, all the neighbourhood without static sensor are excluded. The first uncertainty region is Safe Town as sensor-id 15 had stopped measuring the radiation from 8 April 2020 at around 2300 to 10 April 2020 at around 2000. The radiation in Safe Town should be much higher due to its proximity to the nuclear power plant. Second, the location of sensor-id in some regions like Broadview, Cheddar-ford, and Southwest is not in the centre. Therefore, there is chance that the measurement may not be accurate representation of the radiation level.	Figure 2.5
Mobile Sensor	Easton and East Parton possessed high uncertainty as they are located right beside the Safe Town. Also, being the centre position of St. Himark city, regions such as Weston, Easton, Southton, West Parton, and East Parton has higher uncertainty as the radiation in these regions seem lower as compared to how almost all vehicles pass through these areas.	Figure 2.6
Mobile Sensor	Wilson Forest region should be one of the higher uncertainty areas as there are no radiation measures records in this region. Furthermore, it is located right beside the Wilson Forest Highway which displayed one of the highest radiation records based on Figure 2.7. Last uncertainty area is the Old Town, with the highest concentrated mobile sensors travelled in and out of the area especially it is located near to the Jade Bridge which recorded the highest radiation level on 8 April 2020 at around 2000. /center>	Figure 2.7

c. What effects do you see in the sensor readings after the earthquake and other major events? What effect do these events have on uncertainty?

Sensor Type	Effects after the earthquake or other major events	Evidence
Static Sensor	Referring to Figure 1.3, after the earthquake on 8 April 2020 at around 1500, static sensors-id 12 – 15 logged a rise of radiation from 1600 onwards on the same day. All the static sensors-id continued to log a rise of radiation except for sensor-id 15. Hence, sensor-id 15 contributed to the uncertainty of radiation measurements.	-
Mobile Sensor	On 8 April 2020, mobile sensor-id 10 recorded highest value of radiation. Subsequently, on 9 and 10 April 2020, mobile sensors-id 29, 45, 21, 25, 27, 28, 22, and 24 logged an extreme radiation value and become the outlier of radiation measures. Also, based on the CuSum graph on Figure 1.6, we spot several mobile sensors (29, 45, 21, 25, 27, 28, 22, and 24) recorded extreme radiation level as compared to other mobile sensors. Both figure 1.6 and 2.8 pointed out the same mobile sensors with high radiation measures and therefore the measures is highly certain.	Figure 2.8
Mobile Sensor	Mobile sensor-id 12 recorded the highest radiation measure as compared to other sensors on 9 April 2020 at around 0243 yet the subsequent measures showed a high density of very low radiation value within the healthy range. Hence, it increased the uncertainty of the radiation level.	Figure 2.9