Difference between revisions of "IS428 AY2018-19T1 Lyu Cheng"

From Visual Analytics for Business Intelligence
Jump to navigation Jump to search
(Created page with "<font size="5">'''To be a Visual Detective: Detecting spatio-temporal patterns'''</font> =Overview= In Sofia, Bulgaria, air pollution has been a long-standing serious proble...")
 
 
(14 intermediate revisions by the same user not shown)
Line 1: Line 1:
 
<font size="5">'''To be a Visual Detective: Detecting spatio-temporal patterns'''</font>
 
<font size="5">'''To be a Visual Detective: Detecting spatio-temporal patterns'''</font>
  
 +
=publish link=
 +
https://public.tableau.com/profile/lyu.cheng#!/vizhome/Sofia_Task1/Task-1StoryBoard?publish=yes
 +
https://public.tableau.com/profile/lyu.cheng#!/vizhome/Sofia_Task2/Task2-Story2?publish=yes
 +
https://public.tableau.com/profile/lyu.cheng#!/vizhome/Sofia_Task3/Story3?publish=yes
 
=Overview=
 
=Overview=
  
Line 24: Line 28:
  
 
According to the WHO, 60 per cent of the urban population in Bulgaria is exposed to dangerous (unhealthy) levels of particulate matter (PM10).
 
According to the WHO, 60 per cent of the urban population in Bulgaria is exposed to dangerous (unhealthy) levels of particulate matter (PM10).
−
 
−
=The Data=
 
−
 
−
Official Meteorological Data 
 
−
The official data is used for law suits, policy creation etc. With the far reaching implications, the official data is gathered only from 5 stations, named after neighbourhoods and provides meteorological measurements such as temperature; humidity; pressure etc. This data has longer history, but it’s not spread out across the country. AirBG.info brings to question the quality of this data by suggesting this may have missing data and insufficient measures on the part of the authorities to provide a full representation of Sofia’s air pollution problem.
 
−
 
−
Citizen Meteorological Data
 
−
The Citizen data is gathered from the AirBG.info initiative that is not a government funded and is run by volunteers and citizens of Bulgaria. Each citizen that wishes to participate builds a weather monitoring kit from standardized parts. These citizen weather stations upload data every 5 minutes via an onboard WIFI connectivity and is voluminous in nature. This data has shorter history but is spread across a lot more than 5 stations.
 
−
 
−
In addition it provides data topography data includes Sofia urban area + some areas nominally external to the city (toward the mountains, note large elevation numbers). No particular effort has been made to include entirety of Sofia Capital’s area as per administrative boundaries. This topographical data includes lat/long and elevations for several areas in and around Sofia.
 
−
 
−
Last but not least, the project allows access to API’s that would allow it to gather, inspect and mine data from Citizen Weather station sensors.
 
−
 
−
=Data Quality=
 
−
 
−
In this section, I examine the quality of the data provided by exploring for bad data, gaps in data and informing next steps.
 
−
 
  
 
=Data Cleaning Procedure=
 
=Data Cleaning Procedure=
Line 49: Line 36:
 
| Issue || Bring lat/long/elev data into EEA Data metropolitan data from the metadata.xls file  
 
| Issue || Bring lat/long/elev data into EEA Data metropolitan data from the metadata.xls file  
 
|-
 
|-
−
| Solution || [[File:Dcone.jpg|800px|center]]
+
| Solution || [[File:Dcone.jpg|800px|center]]<br/>
 +
Left merge EEA_Data with metadata.xls.
 
|}
 
|}
  
 
{| class="wikitable"
 
{| class="wikitable"
 
|-
 
|-
−
! Problem #2 || Building Data
+
! Problem #2 || Need consistent aggregation across all data for accuracy.
 
|-
 
|-
−
| Issue || The original data provided, concatenate information into the column itself. For example "F_1_Z_1:Lights Power", it is a column header by itself. The column header tells us that the reading is taken from floor 1, zone 1 and it is measuring the Lights Power. However, such information is only understood by humans instead of business intelligence software like Tableau. Therefore, there is a need for us to transform into a more "software-friendly" form.
+
| Issue || BG_5_60881_2018_timeseries.csv has ‘AveragingTime’ as hour
 
|-
 
|-
−
| Solution || [[Image:Slide2.JPG|800px|center]]
+
| Solution || [[File:Dctwo.jpg|800px|center]]<br/>
 +
 
 
|}
 
|}
  
 
{| class="wikitable"
 
{| class="wikitable"
 
|-
 
|-
−
! Problem #3 || Employee Proximity Card Data
+
! Problem #3 || Goehash cannot be parsed directly by tableau
 
|-
 
|-
−
| Issue || The given images are rich in color to denote the various zones. However, to use it effectively as background for a choropleth map, the image should ideally be dull in color. For example, it should be in colors like gray. Furthermore, the zone boundaries are to be demarcated more obviously especially when it is transformed to color such as gray.
+
| Issue || Geohash is a convenient way of expressing a location (anywhere in the world) using a short alphanumeric string, with greater precision obtained with longer strings, geohash. One geohash value is corresponding to one set of longitude and latitude values. The tableau software needs to use the longitude and latitude values instead of geohash. The data transformation needs to be done.
 
|-
 
|-
−
| Solution || [[Image:Slide3.JPG|800px|center]]
+
| Solution ||  
 +
[[File:Dcthree.png|800px|center]]<br/>
 +
Use coding method to decode all the geohash to long/lat. Notice that the geohash field is still reserved since it is the unique identifier for the different sensors.
 
|}
 
|}
  
 
{| class="wikitable"
 
{| class="wikitable"
 
|-
 
|-
−
! Problem #4 || Employee Proximity Card Data
+
! Problem #4 || Difficulty to identify the data points in the city.
 
|-
 
|-
−
| Issue || The fixed proximity sensor collects data based on the zone which it is in. Unlikely the mobile proximity sensor which bases it detection on coordinates, the fixed sensor detects the cards within its designated zones. Thus, the only information which we are able to get from the fixed sensor is the time which the proximity card is present in the zone. To visualize the zones, we need to mark it out on the image map. Then with the polygon data, I can tell tableau where the zones are on the map. Thus, I need to plot the zones on the given image and retrieve the coordinates. The coordinates are to be saved in a mapping CSV file which will be processed by Tableau.
+
| Issue ||  
 +
In the citizen dataset, the sensor data is across the whole country, while the assignment is mainly focusing on the Sofia city. Data cleaning is required to remove or mark the unrelated data.  
 
|-
 
|-
−
| Solution || [[Image:Slide4.JPG|800px|center]]
+
| Solution ||  
 +
[[File:Dcfour.png|800px|center]]
 +
<br/>
 +
The lat/long boundaries are found in the TOPO-DATA.
 +
Using coding method to compare if the positions of the sensors lie within the city boundary. An additional boolean value is then assigned to each record to indicate whether the sensor is in the country.
 
|}
 
|}
  
 
{| class="wikitable"
 
{| class="wikitable"
 
|-
 
|-
−
! Problem #5 || Employee Proximity Card Data
+
! Problem #4 || pollutant concentration data does not appear in the to meteo data set
 
|-
 
|-
−
| Issue || The given employee data does not provide the prox-id. The proximity data provided by both the fixed and mobile sensors records only prox-id. Those sensors does not capture the name or any other characteristic of the employee. Therefore, there is a need to merge the employee data set and the proximity card data. However, to do so, I need to form the prox-id from the data available from the employee data. After initial observation, the formula for prox-id is as follows : "first name + first letter of last name + 001". However, this formula only works for the majority. There are some which do not obey the formula.
+
| Issue || Merge the concentration data with the meteo data set
 
|-
 
|-
−
| Solution || [[Image:Slide6.JPG|800px|center]]
+
| Solution || Use coding method to align the time format and inner join the two tables.
 
|}
 
|}
  
−
{| class="wikitable"
 
−
|-
 
−
! Problem #6 || Employee Proximity Card Data
 
−
|-
 
−
| Issue || For the visualization, I would need to combine the data logically in Tableau. I would need to combine the files. After the initial combination, I would also need to merge these data with the employee dataset. However, as I am merging multiple files, there have to be common attributes. To ensure that the merge can be successful, I created a row id in the proximity data merged file. Id columns are also created in the individual fixed and mobile proximity files. With the ID, the integration is carried out easily, without confusion to Tableau. As we are trying to clean the data, there would be times where we made modification to the data. Thus, I need a column which would be indepedent of changes/modifications.
 
−
|-
 
−
| Solution || [[Image:Slide7.JPG|800px|center]]
 
−
|}
 
  
−
<b><u>Final Excel Files</u></b>
+
<b><u>Final Data Files</u></b>
−
<ol><li>bldg-MC2.csv</li>
+
<ol><li>pollution_master_data</li>
−
Contains all the building related data (including hazium)
+
This dataset contains the aggragated data of original EEA dataset.
−
<li>bldg-MC2_mapping.csv</li>
+
<li>timeseries</li>
−
Contains all the necessary mapping of the attributes in building data, so that Tableau can understand which floor and zone which the data was taken from.
+
The original EEA dataset
−
<li>employee.csv</li>
+
<li>citizen</li>
−
All the employee related data.
+
The aggragated data of original Air Tube dataset
−
<li>proxData_Merged.csv</li>
+
<li>meteo-concentration</li>
−
Contains the merged data from both fixed and mobile proximity sensor.
+
The aggragated data from the meteo and timeseries data.
−
<li>proxMobileOut-MC2.csv</li>
 
−
Contains all the proximity card data that are recorded by the mobile sensor (Roise).
 
−
<li>proxOUt-MC2.csv</li>
 
−
Contains all the proximity card data that are recorded by the fixed sensor.
 
−
<li>proxOut-MC2_zoning_polygon.csv</li>
 
−
Contains all the polygon mapping of the zones for the fixed proximity sensor.
 
 
</ol>
 
</ol>
−
 
−
=Data Import/Configuration=
 
−
As we are importing multiple files, we need to tell tableau how the files are related to one another. In this case, the files do have a common attribute for all, such as its date/time. However, to allow us to use a filter from one data source to another data source, Tableau needs to understand how the files within the data source are related. For example, in your zone filter, you only want to display zone values which are available at the particular floor which was previously filtered by the user.
 
−
[[Image:Slide 8.jpg|800px|center]]<br>
 
−
<b>Brief Implementation Steps</b><br>
 
−
Once you open up the edit relationship dialog, like the image above. Based on the filter you want to use, choose the common attribute which is present in both datasets. Normally automatic mapping will suffice, however in our case, because of the complexity of our data, Tableau was unable to establish a meaningful relationship between the datasets. Thus, we have to do the custom mapping ourselves.
 
  
 
=Visualisation=
 
=Visualisation=
−
The visualization is based on the category of the data. The breakdown of the proposed visualization is as shown below.
+
Task 1: Spatio-temporal Analysis of Official Air Quality<br/>
−
# Homepage
+
# PM10 Concentration over the timeline
−
# Building Data Explorer : Air Supply Controls / Water Supply Controls / Fan Controls / Coil Controls / Additional System Controls
+
# PM10 Concentration over the timeline with shade
−
# Employee Movement Explorer
+
# PM10 Concentration over Christmas
−
# Variable Explorer
+
Task 2: Spatio-temporal Analysis of Citizen Science Air Quality Measurements<br/>
−
The original dataset is overwhelming. There are over 400 different columns. To make the analysis more meaningful, the data columns has to be group logically based on the purpose of the sensor/data point. I have grouped the data into 6 different categories, namely;
+
# Citizen geo-distribution
−
# Air Supply data
+
# No. of records by hour across the citizen
−
# Water Supply data
+
# Time dependency of sensor data
−
# Fan data
+
Task 3: Relationships between the factors mentioned above and the air quality measure detected in Task 1 and Task 2<br/>
−
# Coil data
+
# Relationship between altitude and concentration
−
# Additional System data
 
−
# Employee Proximity Card data
 
−
 
 
−
The design of the visualization is based on the "Overview first, zoom and filter, then details-on-demand" (Shneiderman, 1996). Thus when the user uses the tool, first he/she will be on the homepage [Step 1]. Through the homepage, it provides an overview of the data exploratory functions available. It provides a summary group of all the available data. The user will then choose the data of his/her interest and be redirected to it.<br>
 
−
 
 
−
Once you are redirected at to the dashboard after [Step 1], you are at [Step 2] now. Basically, at this stage, you are looking at the data which you are interested in. You can interact with the data, by using the filters. Hovering on the data point will provide you more details on the data.<br>
 
−
 
 
−
If you are keen to find out more about the particular dataset/column, you can proceed to [Step 3] where you explore in finer details of the variable which you are looking at. At [Step 3] you proceed to the Variable Explorer. At this dashboard, you are given the ability to drill down the data into finer details. For example, breaking the data up by floor, zones, time etc.<br>
 
  
−
Alternatively, using the Employee Proximity Card data can be either in [Step 2] or [Step 3]. When you are at this dashboard, you can explore the locations of the employees and determine its correlation with the other dataset.
 
  
 
{| class="wikitable"
 
{| class="wikitable"
 
|-
 
|-
−
! style="font-weight: bold;background: #536a87;color:#fbfcfd;width: 20%;" |  [Step 1] Homepage
+
! style="font-weight: bold;background: #536a87;color:#fbfcfd;width: 20%;" |  [Task 1] PM10 Concentration over the timeline
 
|-
 
|-
−
| <b>Purpose / Description</b><br>The homepage is the landing page you will see when you use this Visualisation tool. The data explorery tools are all displayed on the homepage. This homepage makes use of the Tableau Dashboard and its action functions to enable interactivity. It is to serve as a "Home" panel for this visualisation and it would enable the user ease of navigation between the dashboards.
+
| <b>Purpose / Description</b><br>
−
[[Image:Tan Kee Hock MA3 Slide9.JPG|800px|center]]<br>
+
 
 +
This diagram shows the average concentration of the PM10 recorded from the five stations by hours across years.<br/>
 +
 
 +
[[File:oneone.png|800px|center]]
 +
<br>
 
|-
 
|-
 
| <b>Interactive Technique</b><br>
 
| <b>Interactive Technique</b><br>
−
<ol><li>Select : Pointer</li>In order for this homepage to be made possible, there are action rules specified for each of the icons.
+
<ol><li>Select : Pointer</li> A horizontal straight line will be shown once a user clicks on one point on the line, for cross reference over years. The horizontal line is good for direct comparasion with the average line.
−
[[Image:Tan Kee Hock MA3 Slide10.JPG|800px|center]]
 
 
<li>Select : Hover</li>
 
<li>Select : Hover</li>
−
Tooltips are provided to allow the user to understand the action that are tagged to the icon.
+
Tooltips are provided to show air quality station type, averaging tiem, common name, timestamp, average altitude, average concentration.
−
[[Image:Tan Kee Hock MA3 Slide11.JPG|800px|center]]
 
 
</ol>
 
</ol>
 +
<br/>
 +
|-
 +
| <b>Analysis</b><br>
 +
 +
The vertical red drop line indicates the Christmas Days. It is very obvious that the air pollution level grows higher than the other days within one year. This might be mainly because of the fireworks.
 +
 +
Also, a deeper inspection of the data shows, regularly missing data hourly from 9-10 AM from Mladost station (BG0079A) for the critical 1st week of January. The readings in the hours following this missing data spike up significantly. What is the cause of these dropped data signals during these hours? Was there an instrument malfunction in the official weather stations. If the instruments are so costly relative to the citizen weather stations, then is it expected to be unreliable under some conditions.
 +
 +
The missing data from station Orlov and Mladost may cause the average value of the concentration lower than expectation. The maximum concentration among the five stations may be an alternative option, however, that would fail to show the overall situation of the city as the most polluted area is always at the same station.
 +
 +
<br/>
 
|}
 
|}
  
 
{| class="wikitable"
 
{| class="wikitable"
 
|-
 
|-
−
! style="font-weight: bold;background: #536a87;color:#fbfcfd;width: 20%;" | [Step 2] Building Data Explorer : Air Supply Controls / Water Supply Controls / Fan Controls / Coil Controls / Additional System Controls
+
! style="font-weight: bold;background: #536a87;color:#fbfcfd;width: 20%;" | [Task 1] PM10 Concentration over the timeline with shade
 
|-
 
|-
−
| <b>Purpose / Description</b><br> The purpose of this dashboard is to give the user and overview of the data of the related controls. Within the HVAC system, there are a lot of intra-working sub-systems which help keep the entire HVAC system working. This dashboard groups all the related controls together and presents an overview of the data. This will allow the user to easier understand the sub-systems of the data. In general, I had grouped the date into 5 sub-systems:
+
| <b>Purpose / Description</b><br>
−
# Air Supply Sub-System
+
 
−
# Water Supply Sub System
+
This diagram shows the average concentration of the PM10 recorded from the five stations by month across years.<br/>
−
# Fan Sub--System
+
 
−
# Coil Sub-System
+
[[File:Tan Kee Hock MA3 Slide12.JPG|800px|center]]
−
# Additional Sub-Systems
+
<br>
−
The dashboard is logically designed for to ease usability. The layout as shown below.
+
[[File:onetwointeract.png|800px|center]]
−
[[Image:Tan Kee Hock MA3 Slide12.JPG|800px|center]]<br>
+
<br>
−
The dashboard starts with the navigation bar right at the top, followed by the title and description. After which are the filters which are specific for the dashboard. The individual charts then follow. Within the charts itself, it is descriptive by nature. It has its title and this description of what the data is trying to measure.
 
 
|-
 
|-
 
| <b>Interactive Technique</b><br>
 
| <b>Interactive Technique</b><br>
−
<ol>
+
<ol><li>Select : Pointer</li>
 +
The records from a particular station will be highlighted and the rest records become dim.
 
<li>Select : Hover</li>
 
<li>Select : Hover</li>
−
When the user is interested in a specific data point, he/she can simply place the cursor over the data point. A tooltip will instantly appear with the relevant details. This is to provide the user a more granular level of detail.
+
Tooltips are provided to show station name, concentration of PM10, and the timestamp.
−
[[Image:Tan Kee Hock MA3 Slide15.JPG|800px|center]]<br>
 
−
<li>Filter</li>
 
−
The filter at this dashboard is to allow the user to specify the data range at which the user is interested to find more about. For example, he/she wants to look at data that is specific to the month of June and Mondays only. He/She is allowed to do so. At this level, the critical filter is the date. This would give the user an overview of the data within the specific period. Once the filter has been set, the patterns of the individual charts can be seen more clearly.
 
−
[[Image:Tan Kee Hock MA3 Slide13.JPG|800px|center]]<br>
 
−
<li>Connect</li>
 
−
The order of the charts is important. As much as possible, relevant/related charts would be placed adjacent to each other. The scale would be adjusted as well to provide a clearer comparison.
 
−
[[Image:Tan Kee Hock MA3 Slide14.JPG|800px|center]]
 
 
</ol>
 
</ol>
 +
<br/>
 
|-
 
|-
−
| <b>Types of Charts used</b><br>
+
| <b>Analysis</b><br>
−
The data provided are readings taken from various HVAC/Proximity Sensors. Thus, all of the readings are taken against time. To do meaningful comparison and analysis with time as one of the dimension, I used mainly,
+
 
−
# Heatmap
+
A monthly aggregated view shows  Druzhba station having highest peaks during holiday/Christmas times. Druzhba is at 548 meters altitude. This elevation is not very high and a relevant official weather station.
−
# Line Chart
+
 
−
The image below is a representative of the type of charts used. It does not represent all the charts that are present in the dashboard.
+
The missing data from 2017 to 2018 leads to an inaccurate visualisation. According to the previous years, the air pollution level should be lower than what is displayed.
−
[[Image:Tan Kee Hock MA3 Slide16.JPG|800px|center]]
+
 
 +
The changes of the pollution level from the give stations are relative the same. In other words, the concentrations of PM10 from the five stations increase and decrease simultaneously.
 +
<br/>
 
|}
 
|}
  
 
{| class="wikitable"
 
{| class="wikitable"
 
|-
 
|-
−
! style="font-weight: bold;background: #536a87;color:#fbfcfd;width: 20%;" | [Step 2/3] Employee Movement Explorer
+
! style="font-weight: bold;background: #536a87;color:#fbfcfd;width: 20%;" | [Task 1] PM10 Concentration over Christmas
 
|-
 
|-
−
| <b>Purpose / Description</b><br> The purpose of this dashboard is visualise the employee proximity card data. The data are given with X,Y coordinates. Thus, we can plot the data on a background image map which is provided in the original dataset. The proximity card data are visualized on the floor map itself. There are modifications to the floor map so that the data can be better visualized. Now that the employee's movements are visualized on an image map, it gives much higher clarity on the employee's movement/activities around the building.
+
| <b>Purpose / Description</b><br>
−
[[Image:Tan Kee Hock MA3 Slide17.JPG|800px|center]]
+
 
 +
This diagram shows the average concentration of the PM10 recorded from station Hipodruma <br/>
 +
 
 +
[[File:oneThree.JPG|800px|center]]
 +
<br
 
|-
 
|-
 
| <b>Interactive Technique</b><br>
 
| <b>Interactive Technique</b><br>
−
<ol>
+
<ol><li>Select : Hover</li>  
−
<li>Select : Hover</li>
+
Tooltips are provided to show station name, concentration of PM10, and the timestamp.
−
When the user is interested in a specific data point, he/she can simply place the cursor over the data point. A tooltip will instantly appear with the relevant details. This is to provide the user a more granular level of detail.
+
 
−
[[Image:Tan Kee Hock MA3 Slide91.JPG|800px|center]]<br>
 
−
<li>Filter</li>
 
−
The filter at this dashboard is to allow the user to specify the data range at which the user is interested to find more about. Furthermore, the user can filter results based on the employee's department. This allows the user to easily understand the behavior of employees from the different departments. It also helped to show the interaction between each department. Along with the date filters, this would give the user an overview of the employee activity within the specified period. Once the filter has been set, the patterns of the individual charts can be seen more clearly.
 
−
[[Image:Tan Kee Hock MA3 Slide18.JPG|800px|center]]<br>
 
 
</ol>
 
</ol>
 +
<br/>
 
|-
 
|-
−
| <b>Types of Charts used</b><br>For this dashboard, much of the data are given based on the location itself. Thus, the data needs to be plotted on an image to effectively show the pattern between the employee's location and the time of the day. This will help to tell us what the employee's movement/activities are like.
+
| <b>Analysis</b><br>
−
# Bar chart
+
 
−
# Image Maps
+
Christmas period is a typical period that the pollution level will increase dramatically high and reduced to the normal level in 2 days. From the diagram, the concentration increases to 4 times as normal at the afternoon of the 29 Nov. It reaches the highest level at the mid-night; The situation becomes better after the start of 30 Nov.  
−
# Choropleth Map
+
 
−
The image below is a representative of the type of charts used. It does not represent all the charts that are present in the dashboard.
+
<br/>
−
[[Image:Tan Kee Hock MA3 Slide19.JPG|800px|center]]
 
−
[[Image:Tan Kee Hock MA3 Slide20.JPG|800px|center]]
 
 
|}
 
|}
 +
  
 
{| class="wikitable"
 
{| class="wikitable"
 
|-
 
|-
−
! style="font-weight: bold;background: #536a87;color:#fbfcfd;width: 20%;" | [Step 3] Variable Explorer
+
! style="font-weight: bold;background: #536a87;color:#fbfcfd;width: 20%;" | [Task 2] Citizen geo-distribution
 
|-
 
|-
−
| <b>Purpose / Description</b><br> Variable Explorer is to allow the user to further explore the data in more details. In the previous dashboards, especially for the controls, the level of detail is limited so that the analyst can see the bigger picture. In this dashboard, it is designed to empower the analyst to view more about the data and how it changes across floor, zones and time. This is to help the analyst understand how the readings varies across the mentioned building attributes and time. The aim of this dashboard is to focus on just one measurement and understand its pattern/behaviour.
+
| <b>Purpose / Description</b><br>
−
[[Image:Tan Kee Hock MA3 Slide21.JPG|800px|center]]<br>
+
This diagram shows a geospatial distribution of all the sensors across the whole city.
 +
<br/>
 +
[[File:Twoone.png|800px|center]]
 +
<br
 
|-
 
|-
 
| <b>Interactive Technique</b><br>
 
| <b>Interactive Technique</b><br>
−
<ol>
+
<ol><li>Select : Hover</li>  
−
<li>Select : Hover</li>
+
Tooltips are provided to show sensor's latitude/longitude, highest concentration PM10 and hightest concentration PM2.5.
−
When the user is interested in a specific data point, he/she can simply place the cursor over the data point. A tooltip will instantly appear with the relevant details. This is to provide the user a more granular level of detail.
 
−
[[Image:Tan Kee Hock MA3 Slide92.JPG|800px|center]]<br>
 
−
<li>Filter</li>
 
−
There are additional filters to the fundamental date filter. The additional filters are for specifying the measurement variable, floor and zones. The analyst can choose the variable which he/she is interested to find out more about.
 
−
[[Image:Tan Kee Hock MA3 Slide22.JPG|800px|center]]<br>
 
 
</ol>
 
</ol>
 +
<br/>
 
|-
 
|-
−
| <b>Types of Charts used</b><br> The data all have one common attribute, which is date/time. Thus, to enable flexibility for the dashboard to handle all of the variable types, the dashboard is fundamentally be required to visualize time-related data. Therefore, the following types of charts are used.
+
| <b>Analysis</b><br>
−
# Heatmap
+
 
−
# Bar chart
+
This diagram aims to show the geospatial coverage of sensors across the whole country. This is essential since the spatial coverage of the citizen data reflects the confidence and completeness of the whole dataset. This dataset is obtained from citizen database, it is essential to justify the coverage before looking at the pollution level it reflects, if there is some large area is not tracked, the overall result might not be trustworthy.
−
# Line Chart
+
 
−
The image below is a representative of the type of charts used. It does not represent all the charts that are present in the dashboard.
+
Only the data points within the city area are displayed, the irrelevant data is hidden. The way to distinguish the data points is described in the previous data cleaning procedures.
−
[[Image:Tan Kee Hock MA3 Slide23.JPG|800px|center]]<br>
+
 
 +
From the visualization above, the citizen data fairly reflects the overall situation of the country. There is no obvious empty region on the map. However, the North part and the South-East part of the map have a relatively low sensor concentration than the central area. Hence, the pollution records in the central area are more credible.  
 +
 
 +
The colour code is responsible for the highest concentration record reported from the sensor at that location. It can be observed that the points with the deepest colour appear at the centre area indicating that the centre area is the most polluted area.
 +
 
 +
 
 +
<br/>
 
|}
 
|}
  
−
=Use Case=
+
 
 
{| class="wikitable"
 
{| class="wikitable"
 
|-
 
|-
−
! style="font-weight: bold;background: #536a87;color:#fbfcfd;width: 20%;" |  Visualisation Tool Demonstration
+
! style="font-weight: bold;background: #536a87;color:#fbfcfd;width: 20%;" |  [Task 2] No. of records by hour accross citizen
 +
|-
 +
| <b>Purpose / Description</b><br>
 +
 
 +
This diagram shows the number of records reported from the sensors during the past two years <br/>
 +
 
 +
[[File:Twotwo.JPG|800px|center]]
 +
<br
 
|-
 
|-
−
| <b>Scenario</b><br>There is a hardworking analyst who wants to explore for patterns with regards to the bathroom use in the building!
+
| <b>Interactive Technique</b><br>
 +
<ol><li>Select : Hover</li>
 +
Tooltips are provided to show date and the number of records reported at that day.
 +
 
 +
</ol>
 +
<br/>
 
|-
 
|-
−
| <b>Steps</b><br>
+
| <b>Analysis</b><br>
−
[[Image:Tan Kee Hock MA3 Slide24.JPG|800px|center]]<br>
+
 
−
[[Image:Tan Kee Hock MA3 Slide25.JPG|800px|center]]<br>
+
This visualization aims to investigate the time-coverage of the dataset. Over the past two years, the number of records may not be evenly distributed, if there is some period of time what the total records were collected are significantly lower than the rest periods, the data corresponding to this period is not sufficient to showcase the pollution level of the whole country. It also reflects that there were some major failures on the sensors during that period of time.
−
[[Image:Tan Kee Hock MA3 Slide26.JPG|800px|center]]<br>
+
 
−
[[Image:Tan Kee Hock MA3 Slide27.JPG|800px|center]]<br>
+
according to the visual analytics, the records are more concentrated during the second half-year of 2018. This suggests that the sensors' performance was improved at that time. During July 2018, it seems that the sensors report fewer records as compared to other days. Especially in 4th and 5th of July, the data size is approximately ten times lower than other days. The sensors might be under maintenance during that days.
−
[[Image:Tan Kee Hock MA3 Slide28.JPG|800px|center]]<br>
+
<br/>
−
[[Image:Tan Kee Hock MA3 Slide29.JPG|800px|center]]<br>
 
−
[[Image:Tan Kee Hock MA3 Slide30.JPG|800px|center]]<br>
 
−
[[Image:Tan Kee Hock MA3 Slide31.JPG|800px|center]]<br>
 
−
[[Image:Tan Kee Hock MA3 Slide32.JPG|800px|center]]<br>
 
−
[[Image:Tan Kee Hock MA3 Slide33.JPG|800px|center]]<br>
 
−
[[Image:Tan Kee Hock MA3 Slide34.JPG|800px|center]]<br>
 
−
[[Image:Tan Kee Hock MA3 Slide35.JPG|800px|center]]<br>
 
−
[[Image:Tan Kee Hock MA3 Slide36.JPG|800px|center]]<br>
 
−
[[Image:Tan Kee Hock MA3 Slide37.JPG|800px|center]]<br>
 
−
[[Image:Tan Kee Hock MA3 Slide38.JPG|800px|center]]<br>
 
−
[[Image:Tan Kee Hock MA3 Slide39.JPG|800px|center]]<br>
 
 
|}
 
|}
  
−
=Findings  - Task #1=
+
 
−
What are the typical patterns in the prox card data? What does a typical day look like for GAStech employees?
 
 
{| class="wikitable"
 
{| class="wikitable"
 
|-
 
|-
−
! Serial !! Observation
+
! style="font-weight: bold;background: #536a87;color:#fbfcfd;width: 20%;" |  [Task 2] Time dependency of sensor data
 
|-
 
|-
−
| 1 || The people from the facilities department are always around the building 24/7. They are located mostly in level 1. It appears that they work in shifts and they ensure that there is always someone from the department around ay anytime of the day.
+
| <b>Purpose / Description</b><br>
−
[[Image:Tan Kee Hock MA3 Slide41.JPG|800px|center]]
+
 
 +
This diagram shows Time dependency of sensor data <br/>
 +
 
 +
[[File:Twothree.png|800px|center]]
 +
<br>
 +
[[File:Twofour.png|800px|center]]
 +
<br>
 
|-
 
|-
−
| 2 || [[Image:Tan Kee Hock MA3 Slide42.JPG|800px|center]]
+
| <b>Interactive Technique</b><br>
 +
<ol><li>Select : Hover</li>
 +
Tooltips are provided to show date and the concentration.
 +
 
 +
</ol>
 +
<br/>
 
|-
 
|-
−
| 3 || [[Image:Tan Kee Hock MA3 Slide43.JPG|800px|center]]
+
| <b>Analysis</b><br>
−
|-
+
 
−
| 4 || [[Image:Tan Kee Hock MA3 Slide44.JPG|800px|center]]
+
This visualization aims to investigate the time-dependency of the sensor data. If the data shows a common trend across the year, the concentration is time-dependent; if the data fluctuates randomly or keep at a stationary level constantly, it is time-independent.
−
|-
+
 
−
| 5 || [[Image:Tan Kee Hock MA3 Slide45.JPG|800px|center]]
+
The upper diagram shows some random fluctuation due to some anomalies(e.g. PM10=2000), a filter should be implemented to filter out the extreme data.
−
|-
+
 
−
| 6 || [[Image:Tan Kee Hock MA3 Slide46.JPG|800px|center]]
+
The lower diagram is with the filter implemented. From March to August, the pollution concentration level remains at a relative low level. From August to December, it increases and reaches the highest point in January. From January to March the situation becomes better after that and get back to normal level.
−
|-
+
<br/>
−
| 7 || [[Image:Tan Kee Hock MA3 Slide47.JPG|800px|center]]
 
−
|-
 
−
| 8 || [[Image:Slide48.JPG|800px|center]]
 
−
|-
 
−
| 9 || The offices of the employees are arranged by position. The higher position the employee is, it is likely that his/her office will be at the higher floor. The executive departments are mainly located on the 3rd floor, while people from the facility and security department comes from the 1st and 2nd floor.
 
−
|-
 
−
| 10 || Floor 2 is where the bulk of the employees are. Most of the employee's offices are on the 2nd floor. Although their offices are located on the 2nd floor, they still move about the building as frequently. Also, as seen in the floor map and the employee proximity card data, floor 1 is where meetings and front desk offices are located. Thus, the reduced employee presence in floor 1 also suggests that the meeting rooms in floor 1 are likely to be used to host guests/events
 
 
|}
 
|}
  
−
=Findings - Task #2=
 
−
Describe up to ten of the most interesting patterns that appear in the building data. Describe what is notable about the pattern and explain its possible significance.
 
 
{| class="wikitable"
 
{| class="wikitable"
 
|-
 
|-
−
! Serial !! Measurement Category !! Description and Significance
+
! style="font-weight: bold;background: #536a87;color:#fbfcfd;width: 20%;" |  [Task 3] Relationship between concentrtion and altitude
 
|-
 
|-
−
| 1 || Thermostat Setting || The general setting for the thermostat heating and cooling setpoints tend to be opposite of each other. When the heating set point is being set to a higher point, the cooling setpoint will be set to a lower point. This is normally because the user is trying to adjust the temperature of the air within the zone. Naturally, when you want the place to be cooler, you will set the heating point at a lower point, and the cooling point to be at a higher point. This is to produce an equilibrium temperature within the zones. You see that the temperature of the air is between the two setpoints.<br>
+
| <b>Purpose / Description</b><br>
−
[[Image:Tan Kee Hock MA3 Slide50.JPG|800px|center]]<br>
+
 
−
However in the month of June, the period of 7th to 10th. The behavior of the thermostat setting seems to be off the norms. It betrays the general behaviour which is shown in the rest of the month. As the heating setpoint increases, the cooling setpoint increases as well. The general temperature of the air within the zones seems to increase significantly during mid-day. It peaks up as much as to 28.88°C. The average temperature of the air in the zones hovers around 24°C. This is approx. 4°C above the norm. The average temperature in Singapore, especially during the hottest month,February, is around 27°C. The observation here is definitely something worth investigating. The behavior is consistent throughout all the floors and its zone.<br>
+
This diagram shows the relationship between concentration and altitude <br/>
−
[[Image:Tan Kee Hock MA3 Slide51.JPG|800px|center]]<br><br>
+
 
−
There are potential reasoning to this cause.
+
[[File:Threeone.png|800px|center]]
−
# Inappropriate handling of the thermostat controls
+
 
−
# Severe weather conditions - eg Extremely Cool/Hot Weather (Unlikely)
 
−
<b>Significance</b><br>
 
−
The thermostat settings are vital to ensure that the building is properly heated. If the temperature gets too high in the building, and without properly ventilation, will pose potential safety risks to the employees. If the temperature is unable to be regulated induce flavour working conditions, it will likely to cause not just unhappiness but health issues to the employees.
 
 
|-
 
|-
−
| 2|| Mechanical Ventilation Mass Flow Rate || This measurement tells us how much air is flowing through the zone exhaust fan. In the month of June, in particular, there is some inconsistency for the readings on two particular weekends, namely 4th-5th June and 11th-12th June. In general, the readings of this specific measurement has its own cycle within the day. Naturally, it would be lower on the weekends. However, the 2 weekends in June, displays very different reading. The first weekend shows a reading that is below the average while the second weekend shows a reading that is significantly higher than the average.<br>
+
| <b>Interactive Technique</b><br>
−
[[Image:Tan Kee Hock MA3 Slide53.JPG|800px|center]]<br>
+
<ol><li>Select : Hover</li>  
−
You can also observe that the readings are consistent throughout the weekdays and weekends. During the weekday, the flow rate generally increases during mid-day (Possibly due to the hot weather). On the weekend the pattern is very different.<br>
+
Tooltips are provided to show date and the concentration.
−
<b>Significance</b><br>
+
 
−
The readings of the amount of air flowing through the zone exhaust fan can tell us if the building is well ventilated. It indicates the movement of air. Since this observation happens on the weekend, potentially the lack of human activity may be correlated to the lower flow rate. But the difference of flow rate in two separate weekends remains questionable. The flow rate indicates blockage and ventilation of the building. If there is build-up of dust/blockages or animal movement, the flow rate inevitably will be affected. A higher flow rate in the weekend without human activity can potentially indicate faulty sensors or errors in the equipment which results in abnormal control of the ventilation.
+
</ol>
 +
<br/>
 
|-
 
|-
−
| 3 || Bath_Exhaust:Fan Power || This is the measurement of the power used by the bathroom fans. The power indicates usage of the bathroom. There is consistent use of the bathroom throughout the weekday. On the weekend, especially Saturdays (4th and 11th), the usage drops drastically after 1600H.<br>
+
| <b>Analysis</b><br>
−
[[Image:Tan Kee Hock MA3 Slide55.JPG|800px|center]]<br><br>
+
This visualisation aims to investigate the relationship between the altitude and the concentration of pollutants.  
−
<b>Significance</b><br>
+
The five stations located at different altitudes. Among them, the Pavlovo station has the highest altitude while the station Hipodruma is the most polluted station. Hence, there is not a clear relationship between the polltion level and the altitude.
−
The power usage indicates the use of the bathroom. You notice that during the weekday, the bathrooms are consistently used at a similar rate. As explained by the consistent color throughout the working day. This reading tells us the employee's movement and activity of a typical day in the company. The consistent use of bathrooms, indicate human activity in the building as well. Furthermore, it can be used to indicate the employee's productivity, if there are potential cases of "slacking off"/"malingering".
+
 
−
|-
+
 
−
| 4 || Dry Bulb Temperature || The dry-bulb temperature (DBT) is the temperature of air measured by a thermometer freely exposed to the air but shielded from radiation and moisture. DBT is the temperature that is usually thought of as air temperature, and it is the true thermodynamic temperature. Thus, this reading tells us the relative weather condition of outside of the building.<br>
+
<br/>
−
[[Image:Tan Kee Hock MA3 Slide57.JPG|800px|center]]<br>
 
−
As shown in the picture, the readings are very consistent throughout the month of June, you can see that the temperature generally goes up during noon. This reading strongly correlates to the time of the day. Generally, you would expect the temperature to go up during mid-day.<br>
 
−
<b>Significance</b><br>
 
−
The dry bulb temperature is essential for the HVAC system, as the reading can be used to evaluate the effectiveness of the HVAC system within the building. We can  measure how effective the HVAC system is, in regulating the internal building temperature.
 
−
|-
 
−
| 5 || Lights/Pump/Equipment Power|| The readings from all three power consumers, namely lights, pump and equipment display very health power consumption. Their power consumptions are very consistent throughout the month. Light and Equipment power generally peaks up during the weekday. During the weekend, you can see a significant drop in the power consumption. However, for the pump, the power it consumed is a constant number. Either it could be efficiently used, or potentially there is a faulty sensor which causes this reading. Constant reading of 91W can be suspicious.
 
−
[[Image:Tan Kee Hock MA3 Slide94.JPG|800px|center]]<br>
 
−
|-
 
−
| 6 || Water Heater Setpoint & Loop Temp Schedule || The loop temperature schedule refers to the temperature set for the hot water loop. This is the temperature at which hot water is delivered to hot water appliances and fixtures. The temperature for both readings were at a constant value throughout all the month. Both are set at the temperature of 60.0 degree celsius.
 
−
|-
 
−
| 7 || Supply Side Inlet Temperature || This reading measures the temperature of the water entering the hot water tank. The readings intensified as the temperature increases especially on the weekend. The water going into the hot water tank is generally higher during the weekend then compared to the weekday. This is worth investigating as there are lesser human activities over the weekend. The system could be boiling the water unnecessarily, thus, wasting energy.
 
−
[[Image:Tan Kee Hock MA3 Slide95.JPG|800px|center]]<br>
 
−
|-
 
−
| 8 || Lights Power || Despite the consistent total Lights power consumption, there is some interesting pattern to it. Lights power in the first floor is generally not turned off. Much of the power consumption comes from the 1st floor. Even past working hours, the 1st floor still consumes significantly high power, while the rest of the floors' consumption dropped to their minimal level. What is more surprising is that the zones, 8A, 8B, and 11B reflect the lights consumed in corridors. It appears that the building is not really energy efficient after all!
 
−
[[Image:Tan Kee Hock MA3 Slide96.JPG|800px|center]]<br>
 
−
|-
 
−
| 9 || Total Electric Power Demand || The new building claims to be of the highest energy efficiency standards, however, there are questionable data points which do not accurately reflects the energy efficiency capability. The total electric power demand peaks up and intensify on a particular weekend in June (10th - 13th). It begins from Friday morning, and intensify all the way till the following morning. After which, the demand for electric power drops. This is an interesting finding as there should be lower employee activities during the weekends.
 
−
[[Image:Tan Kee Hock MA3 Slide97.JPG|800px|center]]<br>
 
 
|}
 
|}
−
 
−
=Findings - Task #3=
 
−
Describe up to ten notable anomalies or unusual events you see in the data. Prioritize those issues that are most likely to represent a danger or a serious issue for building operations.
 
  
 
{| class="wikitable"
 
{| class="wikitable"
 
|-
 
|-
−
! Priority !! Measurement Category !! Description and Significance
+
! style="font-weight: bold;background: #536a87;color:#fbfcfd;width: 20%;" |  [Task 3] Relationship between concentrtion and temperature
 
|-
 
|-
−
| 1 || Hazium Concentration || Hazium is a recently discovered and possibly dangerous chemical. It poses health hazards to the employees whom inhales it. There are spikes in Hazium concentration especially on 3rd (Friday) and 11th (Saturday) June. What is more surprising is that, one of the areas with high concentration is coming from office 3000(CEO's office).  <br><b>Signifiance</b><br> As mentioned in the background text, hazium is a dangerous chemical. High concentration of haizum is likely to pose health issues to employees. No one can explain the effects of hazium, but it was concluded to likely be a dangerous chemical to employee. Therefore, it is crucial for the company to look for the root course and address it.
+
| <b>Purpose / Description</b><br>
 +
 
 +
This diagram shows the relationship between concentration and temperature<br/>
 +
 
 +
[[File:Threetwo.png|800px|center]]
 +
 
 
|-
 
|-
−
| 2 || Return Outlet CO2 Concentration  || This reading tells us the CO2 concentration within the building. The healthy co2 concentration ranges from 250ppm to 1000pm. However on 2 conservative days (6th and 8th of June), the CO2 concentration spike above 1800 ppm. <br><b>Signifiance</b><br> High concentration of CO2 within the building would post health hazard to the employee. PPM reading above 1000, the employees would experience drowsiness. As it reaches above 2000, employees will experience headaches, sleepiness and stagnant, stale, stuffy air. Poor concentration, loss of attention, increased heart rate and slight nausea. It is vital for the company to look investigate the high CO2 concentration.
+
| <b>Interactive Technique</b><br>
−
[[Image:Tan Kee Hock MA3 Slide93.JPG|800px|center]]
+
<ol><li>Select : Hover</li>
 +
Tooltips are provided to show date and the concentration.
 +
 
 +
</ol>
 +
<br/>
 
|-
 
|-
−
| 3 || Thermostat Setting || This finding is as per one which was mentioned in above in Task #2, the malfunction of this Thermostat would be devastating. <br><b>Signifiance</b><br> The thermostat is responsible for regulating and maintain the internal temperature of the building. You can effectively say that, the readings from the thermostat would control the temperature of the building. There have been instances of it peaking up. The high temperature may potentially cause health hazard for the employee
+
| <b>Analysis</b><br>
−
|-
+
This visualisation aims to investigate the relationship between the temperature and the concentration of pollutants. The relationship is such that the higher the temperature, the lower the pollutant concentration. This might be the cause of the Christmas spike.  
−
| 4 || VAV_SYS Supply Fan Outlet Mass Flow Rate || This reading tells us the total rate of air delivered by the HVAC system fan to the zone it serves. The data collected in the month of June is not showing consistent results.<br>
+
 
−
The readings do tally with the VAV_Sys Supply Fan Outlet:Power.<br>
 
−
[[Image:Tan Kee Hock MA3 Slide59.JPG|800px|center]]
 
−
The readings intensify in 2 particular periods, 7th-8th June and 10th-13th June. During 7th-8th June (Tuesday to Wednesday), the reading intensifies in the early hours and late night. This is an abnormal phenomenon. This is telling us that more air is being delivered by the HVAC system fan when there is no supposed employee during this period. The second period, 10th-13th June, shows intensified readings consistently from 10th June evening to 13th June Morning (Friday to Monday). <br>
 
−
<b>Signifiance</b><br>
 
−
The readings do not seem to tally with the supposed work shifts of employees. There seems to be an increased flow of air during the period where no one supposed to be there. There are multiple possibilities which may have caused such data readings.
 
−
# Faulty Sensors causing false readings (Unlikely)
 
−
# Faulty Equipment
 
−
This reading is important because it will indicate the overall system health of the HVAC fans. It tells us if the HVAC fans are working harder. It also indicates if the HVAC system's ability to maintain the building's internal temperature/ventilation.
 
−
|-
 
−
| 5 || Deli-Fan Power || This reading tells us the power used by the deli exhaust fan. There are some suspicious data points with regards to the use of Deli-Fan.<br>
 
−
[[Image:Slide61.JPG|800px|center]]<br>
 
−
The fan usage seems to be consistently high during a Sunday(5th and 12th June). The readings do not seem to tally with the increased human activities during the weekday. The inconsistent readings do not seem to establish any form of correlation with the human activity. But rather, the pattern of seem to be established by other unknown factors.<br>
 
−
<b>Signifiance</b><br>
 
−
Exhaust fans are health indicators of the overall HVAC systems. Should the exhaust fans power usage display sporadic patterns, they indicate abnormalities within the HVAC system. Furthermore, they help to regulate the airflow for the HVAC system. The poor performance of Exhaust fans will significantly hamper the HVAC's ability to regulate internal building temperature.
 
−
|-
 
−
| 6 || VAV_SYS Heating Coil Power || There is completely 0 power used for the heating coil. This is entirely not possible as the HVAC system seem to be working properly. Thus, there is very little prove that the Heating Coil is broken/faulty.<br>
 
−
[[Image:Tan Kee Hock MA3 Slide62.JPG|800px|center]]<br>
 
−
<b>Signifiance</b><br>
 
−
This is very likely to be a faulty Power Usage sensor. Although this reading does not seem to affect the rest of the system, an investigation in the faulty sensor is recommended. If there are external forces in play which results in the faulty sensor, then it is very likely this cause will impact other parts of the HVAC system. For example, water leakage in a specific part of the building which caused the sensor to be spolit, etc.
 
−
|-
 
−
| 7 || VAV_SYS Supply Fan:Fan Power || The system supply fan consumes more power on the weekend (both Saturday and Sunday). This is highly unusual as there is lower employee activity within the building. Most of the power comes from the fans in level 3. On Saturday it is a half day, but on Sunday only those who are on shift would be in the building. Therefore, on Sunday, there would be close to zero human activity. <br><b>Signifiance</b><br> The supply fan is responsible for circulating the air within the HVAC system. In this case, the unnecessary power consumed by the fan would incur additional cost to the company. Not only that, it is a waste of energy.
 
−
|}
 
  
−
=Findings - Task #4=
 
−
Describe up to five observed relationships between the proximity card data and building data elements. If you find a causal relationship (for example, a building event or condition leading to personnel behavior changes or personnel activity leading to building operations changes), describe your discovered cause and effect, the evidence you found to support it, and your level of confidence in your assessment of the relationship.
 
  
−
{| class="wikitable"
+
<br/>
−
|-
 
−
! Serial || Discovery
 
−
|-
 
−
| 1 ||
 
−
[[Image:Tan Kee Hock MA3 Slide63.JPG|800px|center]]
 
−
[[Image:Tan Kee Hock MA3 Slide64.JPG|800px|center]]
 
−
[[Image:Tan Kee Hock MA3 Slide65.JPG|800px|center]]
 
−
[[Image:Tan Kee Hock MA3 Slide66.JPG|800px|center]]
 
−
<br>
 
−
Thus, the haizum concentration does not occur by chance. It is very likely that someone orchestrated the event. All the clues point towards someone who is likely to be from level 3. More investigation is needed. The attack is very likely to be directed to the CEO himself.
 
 
|}
 
|}
  
−
=Conclusion=
 
−
There are many interesting findings which do not reflect the energy efficiency ability which the builders had claimed to be. The new building does not seem to be as energy-efficient as what was previously advertised. As for the occurrence of Hazium, it is postulated to be caused by the employee themselves. The evidence points towards a deliberate attack towards the CEO himself. As Hazium is a newly discovered chemical, its potential impact on the employees is unknown. Many cautious steps should be taken when investigating the Hazium outbreak. Evident suggest that the culprit seem to be an employee from level 3!<br>
 
−
<b>Main Link</b>
 
−
One tough assignment down, one more project to remaining - https://public.tableau.com/views/MA_3_Final/Home?:embed=y&:display_count=yes
 
−
<br>
 
−
<b>Backup Link</b>
 
−
This is one tough assignment,I need more backup link - https://public.tableau.com/views/MA_3_0/Home?:embed=y&:display_count=yes
 
−
 
−
=Improvement=
 
−
Given more time, i would focus on improving drilling capability of the this visualisation tool. I would also work on improving the interface for the Employee Movement Explorer. But nonetheless, it was a tough fight against time and my analyatical ability. I am still glad manage to generate something like that.
 
 
=Visualisation Software=
 
=Visualisation Software=
−
 
 
To perform the visual analysis, this is a list of the software which I used.
 
To perform the visual analysis, this is a list of the software which I used.
 
*Tableau
 
*Tableau
 
*Excel
 
*Excel
−
*Chrome
+
*VS Code
−
*Netbeans
 
−
 
 
−
=Submission details=
 
−
 
 
−
This is an individual assignment. You are required to work on the assignment and prepare submission individually. Your completed assignment is due on '''24th October 2016, by 12.00 noon'''.
 
−
 
 
−
You need to edit your assignment in the appropriate wiki page of the Assignment Dropbox. The title of the wiki page should be in the form of: IS428_2016-17_T1_Assign3_FullName.
 
−
 
 
−
The assignment 3 wiki page should include the URL link to the web-based interactive data visualization system prepared.
 
−
 
 
−
 
 
−
=Assignment 3 Q&A=
 
  
−
Need more clarification, please feel free to pen down your questions.
 
  
−
#What is Hazium? Hazium is a (fictitious) chemical that has become a recent concern on the island of Kronos. Not much is known about its effects, but it is suspected that Hazium is not good for people.
 
−
#There are a few extra building file data fields in the .json dataset that do not appear in the .csv data. These extra data fields are actually valid for the building for the dates and times they were recorded, but they will not add significantly to your analysis. So for this assignment, please just use the data fields included in the .csv file.
 
−
#Can you provide more info on the data provided in the mobile proximity card data? Are the x,y coordinates bound to a normal (x,y) plane, where in this case the plane is the floor maps? The (x,y) coordinates are bound to a normal plane. The (x,y) plus the floor number would identify a specific location. The lower left of the provided map is (0,0) and the upper right is (189,111).
 
−
#In some cases, data is reported for some sensors and not others, or it is documented but not reported. Where can we find this data? Please use the data fields you have available to perform your investigation. In general, the documented set of attributes may not be reported for all zones.
 
−
#What does the (x,y) coordinates represent for the mobile robot sensor? The (x,y) coordinates for these reading represent the location of the mobile sensor.
 
−
#Sometimes, mobile prox data for a prox card repeats multiple times in a minute. Does this indicate the number of seconds that the prox card was within range of the sensor? No. Multiple readings do not indicate what fraction of the minute that the mobile sensor was in proximity of the prox card.
 
−
#In some cases, the value of the VAV Availability Manager Night Cycle On/Off is 2. Is this a valid value? Yes.
 
−
#Does F_3_Z_9 VAV Damper Position mean F_3_Z_9 VAV REHEAT Damper Position? Yes.
 
  
 
=References=
 
=References=
−
* http://www.picturetopeople.org/image_utilities/image-grayscale-converter/grayscale-image-generator.html
+
* https://www.datasciencesociety.net/sofia-air-quality-eda-exploratory-data-analysis/
−
* https://community.tableau.com/message/320738
+
* https://www.datasciencesociety.net/monthly-challenge-sofia-air-solution-kung-fu-panda/
−
* http://www.thedataschool.co.uk/niccolo-cirone/tableau-tip-week-wednesday-creating-dashboard-navigator-buttons/
+
*  
−
* http://kb.tableau.com/articles/howto/renaming-dimension-column-row-headers
 
−
* https://tableauandbehold.com/2015/04/13/creating-custom-polygons-on-a-background-image/
 
−
* https://www.kane.co.uk/knowledge-centre/what-are-safe-levels-of-co-and-co2-in-rooms
 
−
* https://en.wikipedia.org/wiki/HVAC
 
  
 
=Comments=
 
=Comments=
 
Do provide me your feedback!:)
 
Do provide me your feedback!:)

Latest revision as of 22:53, 11 November 2018

To be a Visual Detective: Detecting spatio-temporal patterns

publish link

https://public.tableau.com/profile/lyu.cheng#!/vizhome/Sofia_Task1/Task-1StoryBoard?publish=yes https://public.tableau.com/profile/lyu.cheng#!/vizhome/Sofia_Task2/Task2-Story2?publish=yes https://public.tableau.com/profile/lyu.cheng#!/vizhome/Sofia_Task3/Story3?publish=yes

Overview

In Sofia, Bulgaria, air pollution has been a long-standing serious problem. Things got so out of control that even the European Court of Justice ruled against Bulgaria in a case brought by the European Commission against the country over its failure to implement measures to reduce air pollution.

Sofia has 5 metropolitan weather stations that capture weather data on hourly intervals. The analysis and comparison are based on the data collected from the five stations. The main measure of pollution is the concentration of a pollutant, PM10. The assignment will explore the factors, for example, humidity, altitude, position, etc., that affect the pollution level.

An interactive and informative visualisation analysis would be designed and developed to demonstrate the result of the result of the above tasks.

The Task

Task 1: Spatio-temporal Analysis of Official Air Quality
Task 2: Spatio-temporal Analysis of Citizen Science Air Quality Measurements
Task 3: Relationships between the factors mentioned above and the air quality measure detected in Task 1 and Task 2

Background Information

Air pollution is an important risk factor for health in Europe and worldwide. A recent review of the global burden of disease showed that it is one of the top ten risk factors for health globally. Worldwide an estimated 7 million people died prematurely because of pollution; in the European Union (EU) 400,000 people suffer a premature death. The Organisation for Economic Cooperation and Development (OECD) predicts that in 2050 outdoor air pollution will be the top cause of environmentally related deaths worldwide. In addition, air pollution has also been classified as a leading environmental cause of cancer.

Air quality in Bulgaria is a big concern: measurements show that citizens all over the country breathe in air that is considered harmful to health. For example, concentrations of PM2.5 and PM10 are much higher than what the EU and the World Health Organization (WHO) have set to protect health.

Bulgaria had the highest PM2.5 concentrations of all EU-28 member states in urban areas over a three-year average. For PM10, Bulgaria is also leading on the top polluted countries with 77 μg/m3on the daily mean concentration (EU limit value is 50 μg/m3).

According to the WHO, 60 per cent of the urban population in Bulgaria is exposed to dangerous (unhealthy) levels of particulate matter (PM10).

Data Cleaning Procedure

Problem #1 Location is needed for final result to be shown as map and is a learning feature for NN
Issue Bring lat/long/elev data into EEA Data metropolitan data from the metadata.xls file
Solution
Dcone.jpg

Left merge EEA_Data with metadata.xls.

Problem #2 Need consistent aggregation across all data for accuracy.
Issue BG_5_60881_2018_timeseries.csv has ‘AveragingTime’ as hour
Solution
Dctwo.jpg

Problem #3 Goehash cannot be parsed directly by tableau
Issue Geohash is a convenient way of expressing a location (anywhere in the world) using a short alphanumeric string, with greater precision obtained with longer strings, geohash. One geohash value is corresponding to one set of longitude and latitude values. The tableau software needs to use the longitude and latitude values instead of geohash. The data transformation needs to be done.
Solution
Dcthree.png

Use coding method to decode all the geohash to long/lat. Notice that the geohash field is still reserved since it is the unique identifier for the different sensors.

Problem #4 Difficulty to identify the data points in the city.
Issue

In the citizen dataset, the sensor data is across the whole country, while the assignment is mainly focusing on the Sofia city. Data cleaning is required to remove or mark the unrelated data.

Solution
Dcfour.png


The lat/long boundaries are found in the TOPO-DATA. Using coding method to compare if the positions of the sensors lie within the city boundary. An additional boolean value is then assigned to each record to indicate whether the sensor is in the country.

Problem #4 pollutant concentration data does not appear in the to meteo data set
Issue Merge the concentration data with the meteo data set
Solution Use coding method to align the time format and inner join the two tables.


Final Data Files

  1. pollution_master_data
    2. This dataset contains the aggragated data of original EEA dataset.
  2. timeseries
    3. The original EEA dataset
  3. citizen
    4. The aggragated data of original Air Tube dataset
  4. meteo-concentration
    5. The aggragated data from the meteo and timeseries data.

Visualisation

Task 1: Spatio-temporal Analysis of Official Air Quality

  1. PM10 Concentration over the timeline
  2. PM10 Concentration over the timeline with shade
  3. PM10 Concentration over Christmas

Task 2: Spatio-temporal Analysis of Citizen Science Air Quality Measurements

  1. Citizen geo-distribution
  2. No. of records by hour across the citizen
  3. Time dependency of sensor data

Task 3: Relationships between the factors mentioned above and the air quality measure detected in Task 1 and Task 2

  1. Relationship between altitude and concentration


[Task 1] PM10 Concentration over the timeline
Purpose / Description

This diagram shows the average concentration of the PM10 recorded from the five stations by hours across years.

Oneone.png


Interactive Technique
  1. Select : Pointer
    2. A horizontal straight line will be shown once a user clicks on one point on the line, for cross reference over years. The horizontal line is good for direct comparasion with the average line.
  2. Select : Hover
    3. Tooltips are provided to show air quality station type, averaging tiem, common name, timestamp, average altitude, average concentration.


Analysis

The vertical red drop line indicates the Christmas Days. It is very obvious that the air pollution level grows higher than the other days within one year. This might be mainly because of the fireworks.

Also, a deeper inspection of the data shows, regularly missing data hourly from 9-10 AM from Mladost station (BG0079A) for the critical 1st week of January. The readings in the hours following this missing data spike up significantly. What is the cause of these dropped data signals during these hours? Was there an instrument malfunction in the official weather stations. If the instruments are so costly relative to the citizen weather stations, then is it expected to be unreliable under some conditions.

The missing data from station Orlov and Mladost may cause the average value of the concentration lower than expectation. The maximum concentration among the five stations may be an alternative option, however, that would fail to show the overall situation of the city as the most polluted area is always at the same station.


[Task 1] PM10 Concentration over the timeline with shade
Purpose / Description

This diagram shows the average concentration of the PM10 recorded from the five stations by month across years.

Tan Kee Hock MA3 Slide12.JPG


Onetwointeract.png


Interactive Technique
  1. Select : Pointer
    2. The records from a particular station will be highlighted and the rest records become dim.
  2. Select : Hover
    3. Tooltips are provided to show station name, concentration of PM10, and the timestamp.


Analysis

A monthly aggregated view shows Druzhba station having highest peaks during holiday/Christmas times. Druzhba is at 548 meters altitude. This elevation is not very high and a relevant official weather station.

The missing data from 2017 to 2018 leads to an inaccurate visualisation. According to the previous years, the air pollution level should be lower than what is displayed.

The changes of the pollution level from the give stations are relative the same. In other words, the concentrations of PM10 from the five stations increase and decrease simultaneously.

[Task 1] PM10 Concentration over Christmas
Purpose / Description

This diagram shows the average concentration of the PM10 recorded from station Hipodruma

OneThree.JPG

<br

Interactive Technique
  1. Select : Hover
    2. Tooltips are provided to show station name, concentration of PM10, and the timestamp.


Analysis

Christmas period is a typical period that the pollution level will increase dramatically high and reduced to the normal level in 2 days. From the diagram, the concentration increases to 4 times as normal at the afternoon of the 29 Nov. It reaches the highest level at the mid-night; The situation becomes better after the start of 30 Nov.



[Task 2] Citizen geo-distribution
Purpose / Description

This diagram shows a geospatial distribution of all the sensors across the whole city.

Twoone.png

<br

Interactive Technique
  1. Select : Hover
    2. Tooltips are provided to show sensor's latitude/longitude, highest concentration PM10 and hightest concentration PM2.5.


Analysis

This diagram aims to show the geospatial coverage of sensors across the whole country. This is essential since the spatial coverage of the citizen data reflects the confidence and completeness of the whole dataset. This dataset is obtained from citizen database, it is essential to justify the coverage before looking at the pollution level it reflects, if there is some large area is not tracked, the overall result might not be trustworthy.

Only the data points within the city area are displayed, the irrelevant data is hidden. The way to distinguish the data points is described in the previous data cleaning procedures.

From the visualization above, the citizen data fairly reflects the overall situation of the country. There is no obvious empty region on the map. However, the North part and the South-East part of the map have a relatively low sensor concentration than the central area. Hence, the pollution records in the central area are more credible.

The colour code is responsible for the highest concentration record reported from the sensor at that location. It can be observed that the points with the deepest colour appear at the centre area indicating that the centre area is the most polluted area.




[Task 2] No. of records by hour accross citizen
Purpose / Description

This diagram shows the number of records reported from the sensors during the past two years

Twotwo.JPG

<br

Interactive Technique
  1. Select : Hover
    2. Tooltips are provided to show date and the number of records reported at that day.


Analysis

This visualization aims to investigate the time-coverage of the dataset. Over the past two years, the number of records may not be evenly distributed, if there is some period of time what the total records were collected are significantly lower than the rest periods, the data corresponding to this period is not sufficient to showcase the pollution level of the whole country. It also reflects that there were some major failures on the sensors during that period of time.

according to the visual analytics, the records are more concentrated during the second half-year of 2018. This suggests that the sensors' performance was improved at that time. During July 2018, it seems that the sensors report fewer records as compared to other days. Especially in 4th and 5th of July, the data size is approximately ten times lower than other days. The sensors might be under maintenance during that days.


[Task 2] Time dependency of sensor data
Purpose / Description

This diagram shows Time dependency of sensor data

Twothree.png


Twofour.png


Interactive Technique
  1. Select : Hover
    2. Tooltips are provided to show date and the concentration.


Analysis

This visualization aims to investigate the time-dependency of the sensor data. If the data shows a common trend across the year, the concentration is time-dependent; if the data fluctuates randomly or keep at a stationary level constantly, it is time-independent.

The upper diagram shows some random fluctuation due to some anomalies(e.g. PM10=2000), a filter should be implemented to filter out the extreme data.

The lower diagram is with the filter implemented. From March to August, the pollution concentration level remains at a relative low level. From August to December, it increases and reaches the highest point in January. From January to March the situation becomes better after that and get back to normal level.

[Task 3] Relationship between concentrtion and altitude
Purpose / Description

This diagram shows the relationship between concentration and altitude

Threeone.png
Interactive Technique
  1. Select : Hover
    2. Tooltips are provided to show date and the concentration.


Analysis

This visualisation aims to investigate the relationship between the altitude and the concentration of pollutants. The five stations located at different altitudes. Among them, the Pavlovo station has the highest altitude while the station Hipodruma is the most polluted station. Hence, there is not a clear relationship between the polltion level and the altitude.



[Task 3] Relationship between concentrtion and temperature
Purpose / Description

This diagram shows the relationship between concentration and temperature

Threetwo.png
Interactive Technique
  1. Select : Hover
    2. Tooltips are provided to show date and the concentration.


Analysis

This visualisation aims to investigate the relationship between the temperature and the concentration of pollutants. The relationship is such that the higher the temperature, the lower the pollutant concentration. This might be the cause of the Christmas spike.



Visualisation Software

To perform the visual analysis, this is a list of the software which I used.

  • Tableau
  • Excel
  • VS Code


References

Comments

Do provide me your feedback!:)

Retrieved from ‘https://wiki.smu.edu.sg/18191is428g1/index.php?title=IS428_AY2018-19T1_Lyu_Cheng&oldid=2926’