1 - Critique the visualization
Figure 1: The Original Visualization (with Markups)
1.1 Clarity
| S/N | Comments | Suggested Improvements |
|---|---|---|
| C1 | Points are plotted all across the map and it can hard to get a sense of severity or the number of incidents in each area. | Scale the points by size (e.g. Number of casualties) and make them transparent so underlying points can be seen. (Proportion Symbol Map). |
| C2 | The line-plots all have different vertical scales which can make it hard to compare event counts across different categories. | Explore plotting them all on one graph on the same axis with color coding. This allows users to compare the counts across different event types as well. Doing this will free up space to add another chart. Fatalities by event type, but converted to a logarithmic scale |
| C3 | Horizontal axis labeled as ‘Event Date’ but events are plotted on a ‘Year’ axis. | Change axis label to ‘Event Year’ |
| C4 | The vertical axis label ‘Count of Sheet1’ does not convey any meaningful message | Replace the label with ‘Event Count’ |
1.2 Aesthetics
| S/N | Comments | Suggested Improvements |
|---|---|---|
| A1 | No vertical gridlines on line plots. This makes it hard to read the plots when they all share the horizontal axis at the bottom. | Include vertical gridlines to allow users to clearly identify the year in which the points are plotted |
| A2 | Points are only colour coded and hard to tell apart when overlapping and scrunched together. | Points should also be scaled by size and translucent so that overlapping points are not lost |
| A3 | Colours on the legend are not consistent throughout the dashboard. | For consistency sake, the line plots should have the same colour as the points on the map. |
1.3 Interactivity
| S/N | Comments | Suggested Improvements |
|---|---|---|
| I1 | The map visualization is re-centered whenever the dashboard is filtered by another country. This allows the user to quickly focus on events and locations in said country. However, it can be hard to identify events that spill-over from neighbouring countries or along borders. | Change filter to allow multiple selections. Continue using the function where map re-centers around the filtered area. |
| I2 | There are a lot of data points that are unused from the dataset that can be plotted on the map such as ‘sub-event type’ or ‘actors’ | Create a tooltip to incorporate these data points. |
| I3 | There is no way to filter by ‘Event date’ this could be important information to see if events on the map are changing over time. | Add in filtering by ‘event date’ |
2 - The suggested visualization
The proposed design is as follows.
Figure 2: The proposed data visualization with improvements)
3 - The Visualization process
The Data Source used is from the The Armed Conflict Location & Event Data Project (ACLED). The data can be found here.
3.1 Steps taken for first visualization (Map)
Download data file with an .xlsx file format.
Load file into Tableau Desktop
Check and ensure that data-types for geographical information are correctly identified.
Figure 3: Ensuring Lat/Long attributes have geographical roles)
- Drag the longitude and latitude fields on to the column and row shelves respectively.
Figure 4: Creating the map
- Drag Event Type and Event Id Cnty to the Marks card.
Figure 5: Plotting the points
- Drag Fatalities to the size and make the measure a SUM. Increase the size of the marks
Figure 6: Scaling the size of points by fatalities
- Change the range of the size legend from 0 - 243 to 0 - 250 to have consistent ranges.
Figure 7: Creating a consistent legend
Rename the title of the size legend from SUM(Fatalities) to Number of Fatalities
Decrease the opacity of the circles in the colour menu from 100% to 50% to allow overlapping points to show.
Figure 8: Making the circles partially transparent
- Drag the Event Date on to the filter pane. Select Years as filter method and select all for filtering.
Figure 9: Creating a Year filter for events
Drag Country on to the filter pane. Select all countries for filtering.
Drag Event Type on to the filter pane. Select all types for filtering.
Drag Sub Event Type on to the filter pane. Select all types for filtering.
Drag Admin1 on to the filter pane. Select all types for filtering.
Figure 10: Adding all the other filters
- Show the Country, Admin1, Event Type and Sub Event Type filters . Change filter type to Multiple Values (Dropdown) in order to better manage space in the dashboard.
Figure 11: Showing and decluttering the filters created
- For Admin1, change the display criteria to ‘Only Relevant Values’. This will make it only show data points relevant to the selection of other filters. e.g. only show Admin1 values relevant to Cambodia when country is filtered by Cambodia.
Figure 12: Adding the State filter and only showing values to selected country
Rename the filter Admin1 to State
Show the YEAR (Event Date) filter.
Figure 13: Exposing filters to users
- Include Sub Event Type, Actor 1, Assoc Actor 1, Actor 2, Assoc Actor 2, Event Date, Country, Admin1 into the tooltips
Figure 14: Adding more information into tooltips
- Reformat the tooltip into three distinct parts ‘Event Details’, ‘Event Location’ and ‘Involved Parties’
Figure 15: Reformatting the tooltips
- Rename the visualization to Armed Conflict Location & Event in < Country>, <YEAR(Event Date)>
Figure 16: Renaming the visualization
3.2 Steps taken for second visualization (Line Graph)
Create a new sheet.
Drag Sheet1(COUNT) to the rows shelf
Figure 17: Populating the graph
- Drag event type on to colour on the marks card.
Figure 18: Color coding the graph
- Drag Year to the columns shelf
Figure 19: Plotting the line plots against year
Rename vertical axis as Number of Events
Create a new sheet Event Count Tooltip
Drag Sheet1(COUNT) to the columns shelf and the marks card.
Drag Country to the rows shelf
Figure 20: Populating the data points for the tooltip
- Rename the horizontal axis as Number of Events
Figure 21: Renaming the axis
- Add in sheet as tooltip in the original visualization. Tooltip will show breakdown of event counts by country for each given event type.
Figure 22: How the tooltip and formatting looks like
- Rename chart as Event Count
Figure 23: Renaming the chart
3.3 Steps taken for third visualization (Line Graph)
- Duplicate the previous sheet for event counts.
Figure 24: Duplicating the previous visualization
- Replace Sheet1(COUNT) with SUM(Fatalities) on the rows shelf
Figure 25: Replacing attributes: Count of events with sum of fatalities
- Rename vertical axis as Fatalities’ and change to logarithmic to prevent the graph from being distorted
Figure 26: Renaming the axis and changing the scale to log
Create a new sheet.
Drag Fatalities to the columns shelf and the marks card and Drag Country to the rows shelf
Figure 27: Creating a second visual tooltip
- Add in sheet as tooltip in the original visualization. Tooltip will show breakdown of fatalities by country for each given event type.
Figure 28: How the tooltip and formatting looks like
- Rename chart as Number of Fatalities
Figure 29: Renaming the chart
3.4 Steps taken for fourth visualization (Data Table)
- Drag Sheet1(COUNT) to text on the marks card
Figure 30: Creating a data table
- Drag country to the rows shelf
Figure 31: Breaking down the data table by country
- Rename visualization to Events by Country
Figure 32: Renaming the chart
- Hide field labels for rows
Figure 33: Removing redundant field label
3.5 Steps taken to create dashboard
Create new dashboard tab
Change the size of the dashboard from fixed to automatic
Figure 34: Changing dashboard size setting
- Drag the map on to the dashboard
Figure 35: Putting the map visual on the dashboard
- Drag the event count visualization to the right of the map
Figure 36: Putting the event count visual on the dashboard
- Place the fatalities visualization to the bottom of the event count visualization
Figure 37: Putting the fatalities visual on the dashboard
- Drag the events by country data table to the bottom right corner.
Figure 38: Putting the event count data table on the dashboard
- Synchronize filters to apply across all visualizations
Figure 39: Changing filters to apply across all visuals
- Dashboard is complete.
4 - The improved data visualization
The completed visualization is as follows.
Figure 40: The Completed dashboard
Figure 41: The Completed dashboard with a visual tooltip on fatalities displayed
Figure 42: The Completed dashboard with a visual tooltip on event count displayed
The interactive dashboard is uploaded onto Tableau Public Server and can be found here.
5 - New Insights
5.1 Myanmar, Philipines and Thailand are the deadliest countries
Figure 43: Location of the most deadly events
As seen by the diagram, Myanmar, Philipines and Thailand are indicated by the blue arrows. These are areas in which there are the highest concentration of circles and more importantly - big circles. This means that events in these countries usually result in a huge loss of life.
Comparatively, countries like Vietnam have relatively minor events compared to these countries.
5.2 The single deadliest event happened in Myanmar
Figure 44: The deadliest event
The single deadliest event happened in the Rakhine state in on 28 August 2017. On that day 243 people died in a clash between the Myanmar military and the Rohingya muslims. The violence against the Rohingya muslims are well known and commonplace. In fact, almost all of the violence against civilian events in the Rakhine state were carried out against the Rohingya civilian muslims.
5.3 Indonesia: Many protests but protests are barely fatal.
Figure 45: Indonesia’s Events over the years and locations
As can be seen in the event counts line graph: Protests are by far the most common events in Indonesia and the number of protests have grown considerably in the second half of the decade and even surpassing 1000 in 2020 (as seen by the orange arrows). Deaths associated with protest events however are always considerably rare. The deadliest year of protests only resulted in 5 deaths (2015). On good years, protests have resulted in zero deaths - including the peak year of 2020.
From the map, we can also observe an intersting sign - Kalimantan has least events and fatalities (blue arrow). Compared to the rest of the states like Papua, Java, Aceh and Sulawesi (red arrows), Kalimantan rarely has armed conflicts (as seen by the relatively fewer number of circles) and even if there are events, they usually do not result in massive loss of lives (circles present in Kalimantan are small).
5.4 WMD are used only in a few countries
As identified in 5.1, Thailand, Mynamar and Philipines are the countries where most deaths occur. This could be partly due to the use of explosives/remote violence in these events. Looking at the sub-event types (called out in the dropdown list), these events usually involve weapons that can harm many people at once and can be termed as what we call weapons of mass destruction (WMD).
Figure 46: WMD events over the years and locations
These events are usually conducted by 2 main actors.
The state against civilians. The listed countries generally do not have any qualms against using military grade equipment such as air or artillery strikes against civilians. These usually results in huge loss of lives as civilians barely have any power to retaliate.
Armed militia groups or terror groups. There are groups within these countries with access to sophisticated weaponry (think lax gun control laws). Such groups can range from Abu Sayyaf in the Philippines to the JAD in Indonesia to the Malay Muslim Separatist in the Thai border near Malaysia.
5.5 Spike in 2016 deaths and events caused by data from Philipines
Figure 47: Correlating numbers to Phillipines data
Figure 48: Phillipines spike in 2016
Data from Phillipines was not made available until 2016. It was also that year where Philipines almost singlehandedly contributed to the entire death count caused by violence to civilians and battles. This could be attributed to President Duerte taking office in June 2016 and declaring a drug war. The violence to civilians could be collateral damage and battles could be with drug gangs resisting against the crackdown.
This could mean that the number of events and fatalities before 2016 could be even higher had data from Philippines been made available. Following the addition of Phillipines to the dataset, deaths due to violence towards civilians and battles from Phillipines remained the highest until 2019 where the conflict between the Rohingya muslims and the Myanmar government intensified in the Rakhine state.
Figure 49: Numerous deadly events in Rakhine in 2019