Skip To Content

Interpret patterns using link analysis

In the previous lesson, you created nonspatial cards and filters. In this lesson, you'll use the maps, filters, charts, and tables to look for patterns in your data. You'll also interpret those patterns by performing link analysis and looking for discrepancies between situations with similar types of conflicts and changes over time.

Analyze current refugee situations

In 2017, the countries of origin with the highest total persons of concern were Syria and Colombia. You'll use the charts created in the previous section to find out more about the situations in those countries.

  1. If necessary, open your UNHCR Population Statistics workbook.
  2. For the predefined filter, make sure 2017 is chosen.
  3. On the column chart, select the Syria column. Zoom and pan to Syria on the map. If you need help finding Syria, look for the area where all of the links are originating.
    Tip:

    Use the scroll bar on your mouse or use Shift and draw a box around Syria on the map to zoom. If you click on selected features when you pan or draw the box, the features will be dragged when you move your mouse. This functionality is one of the drag and drop features in Insights that can be used to create new maps, charts, and tables from the selections on your cards. Therefore, you will have to click your mouse over an area of the map that does not have selected lines or nodes on it.

    The bubble chart and summary table are both filtered to show the countries where persons of concern from Syria were residing in 2017. You can also click the arrows next to the column name in the summary table to sort the columns ascending or descending.

    According to the calculations at the bottom of the table, there are over 6.3 million refugees, 147,000 asylum seekers, and 6.15 million internally displaced persons from Syria, totaling over 13 million persons of concern. There are other categories of persons of concern in the UNHCR data that are not present on the table, including returned refugees, returned internally displaced persons, stateless persons, and others of concern, which is why the sum of the total population is slightly larger than the sum of refugees, asylum seekers, and internally displaced persons. The majority of the data is in the refugees, asylum seekers, and internally displaced persons fields, so you'll only be using those fields in the table for now.

    On the map, the thickness of the links between Syria and the country of residence indicate the number of persons of concern moving between those countries. The thickest line is the loop back to Syria, which represents mostly internally displaced persons. The next thickest lines are to Turkey, Lebanon, and Jordan. These three lines represent approximately 80 percent of the refugees from Syria.

    Persons of concern originating in Syria

    The flow map uses the Total Population field, so how does Insights know how many persons of concern there are internally in Syria? The flow map pulls data from the Total Population field, visualized as line widths, and links it to country of origin and country of residence locations. However, displaying the data spatially gives more context and information that helps you to understand the data better.

    Turkey, Lebanon, and Jordan all share a border with Syria, which indicates that proximity plays a major role in where refugees are residing. The other neighboring countries are Iraq and Israel; according to the summary table, Iraq had the fifth largest number of refugees in 2017 behind Germany, but Israel only had two Syrian refugees, despite the proximity of the two countries. Unfortunately, the data cannot tell you why the number of refugees in Israel is so much smaller than in the other neighboring countries. You'll have to do more research if you want to find out why.

    Besides giving information about the number of migrants between each country, the flow map also has information about how many countries the refugees are coming from through the centrality in the nodes.

  4. Point to the node for Turkey.

    Normalized indegree of Turkey

    A pop-up appears and shows that Turkey has an indegree centrality of 0.37. The centrality numbers are normalized, meaning they are showing the proportion of countries of origin to which each country of residence is connected. There are a lot of reasons to use normalized centrality values, such as comparing values across networks with different sizes. In this example, normalization is not necessary and not intuitive. You'll turn off normalization on the nodes.

  5. On the map card, for the Persons of Concern.Table1 layer, expand the Layer options pane. In the Options tab, uncheck Normalized.

    Normalized parameter

  6. Close the Layer options pane.
  7. Point to the nodes for Turkey, Lebanon, Jordan, and Germany.
    Tip:

    You may need to pan or zoom the map to see the node for Germany.

    Indegree centrality of Turkey

    The indegree centralities are now 77, 62, 54, and 164, respectively.

  8. On the column chart, deselect Syria.

    Colombia has the second highest number of persons of concern, according to both the bubble chart and the column chart. You will repeat the analysis for Colombia.

  9. On the Country of origin chart, select Colombia (the second highest column).

    Colombia selected on the column chart

    The map is filtered to show Colombia as the country of origin. Zoom and pan to northern South America and center your map around Colombia. Unlike Syria, it is difficult to tell which lines are thickest because the number of internally displaced persons is so much larger than the number of refugees in other countries of residence. You'll have to rely on the summary table to find the countries with the most refugees from Colombia.

  10. On the summary table, next to Total Population and Sum, click the arrows twice.

    The column is sorted in descending order, with the countries that have the highest number of persons of concern shown first. Colombia has almost 7.7 million internally displaced persons and over 190,000 refugees. The countries with the most refugees from Colombia are Venezuela, Ecuador, the United States, and Canada.

    Venezuela and Ecuador both neighbor Colombia, but the United States and Canada do not. However, the number of refugees in the United States and Canada is much lower than the number in Venezuela and Ecuador, so the pattern of refugees residing mostly in neighboring countries that you saw with Syria is still consistent for Colombia.

  11. Point to the nodes for Venezuela, Ecuador, the United States, and Canada.
    Tip:

    You can use the Default Extent button to zoom to the map's original extent, rather than panning to the United States and Canada.

    The Default Extent button

    The indegree centrality values for Venezuela, Ecuador, the United States, and Canada are 28, 87, 185, and 178, respectively. Based on the other information in the node pop-up, the maximum centrality in 2017 was 185, meaning the United States had the highest indegree centrality of any country.

    For both Syria and Colombia, the country of origin had the largest number of persons of concern, which indicates the high number of internally displaced persons.

    Another pattern seen in both countries is that persons of concern tend to travel the shortest distance to find refuge, such as Syrians finding refuge in the neighboring countries of Turkey, Lebanon, and Jordan. The exceptions are countries such as the United States, Canada, and Germany, which are countries known to have large immigrant and refugee populations. These countries, therefore, have higher indegree centralities than countries who mostly take in persons of concern from neighboring countries, despite often taking in lower numbers of persons of concern.

    These trends do not explain the cause of the human flows, but they serve as a starting point for further investigation. Global affairs, war, politics, foreign policies, and other factors all influence what boundaries people are allowed to cross.

Analyze similar refugee situations

In 2017, the United Nations High Commissioner for Human Rights called the violence against Rohingya Muslims in Myanmar a "textbook example of ethnic cleansing." Next, you'll look closer at the data for Myanmar and compare it to past conflicts involving ethnic cleansing to determine if any patterns emerge among similar types of conflicts.

You'll analyze data from the current conflict in Myanmar and the Rwandan genocide. Because you already know what data you will be analyzing, you can create a new predefined filter to make it easier to switch between the two countries.

  1. If necessary, deselect any selections on the column chart.
  2. In the data pane, click the Widget drop-down arrow and choose Predefined Filter.

    The Predefined Filter 1 card is added to the page. You may need to scroll down the page to see the new filter.

  3. On Predefined Filter 1, click Add.

    The New Filter pane appears.

  4. For Filter by, under Persons of Concern.Table1, choose Origin.

    Filter by Origin field

  5. Uncheck Select All.

    Uncheck Select All

    All of the countries are deselected.

  6. In the search bar, type Myanmar. In the list of results, check Myanmar.

    Myanmar search result

  7. Search for Rwanda. In the list of results, check Rwanda.
  8. Click By value.

    The two selected values are added to the filter.

    Origin filter

  9. Rename Predefined Filter 1 to Origin. If necessary, resize the card and move it to a convenient place on your page.
  10. On the Origin filter, check Myanmar.

    Unlike in the previous section, the largest group of persons of concern who left Myanmar is refugees. The second largest group is internally displaced persons. In the other countries you looked at, the largest group was internally displaced persons.

    Table of persons of concern from Myanmar

    Based on the flow map, the refugees from Myanmar seem to reside in fewer countries than refugees from the previous countries you looked at. You'll quantify the pattern by changing the centrality of the nodes from Indegree to Outdegree.

  11. On the Migration of persons of concern map card, for the Persons of Concern.Table1 layer, expand the Layer options pane. In the Options tab, change Indegree to Outdegree.

    Display the nodes by outdegree

    Until now, the Migration of persons of concern map did not have a filter applied because filtering the data would affect the indegree calculations. Now that the centrality has been changed to Outdegree, the centrality of the origin nodes will not be affected when filtered by the country of origin.

  12. Close the Layer options pane. Zoom to Myanmar and point to the Myanmar node.

    Outdegree centrality for Myanmar

    The outdegree for Myanmar is 52, meaning persons of concern from Myanmar were residing in 52 countries in 2017.

    Next, you'll look at the years preceding the genocide in Myanmar to see if there are any patterns in the outdegree centrality or in the number of persons of concern leading up to the conflict.

  13. On the Year filter, click 2015.
  14. On the Migration of persons of concern map, point to the node for Myanmar.

    The outdegree is 51. The Refugees, asylum seekers, and internally displaced persons table indicates that there are 451,805 refugees, 60,616 asylum seekers, 451,089 internally displaced persons, and 989,220 persons of concern in total.

  15. Change the Year filter to 2016.

    The following table summarizes the data for the specified years:

    YearOutdegree centralityRefugeesAsylum seekersInternally displaced personsTotal population

    2015

    51

    451,805

    60,616

    451,089

    989,220

    2016

    51

    490,289

    55,976

    375,016

    922,963

    2017

    52

    1,156,746

    41,092

    353,108

    1,551,740

    There is only one country difference between the outdegree in 2015 and 2016 compared with 2017, despite an increase in the number of refugees by over 600,000 persons. The number of internally displaced persons decreased slightly each year.

  16. Use the Year filter and the Refugees, asylum seekers, and internally displaced persons table to compare the number of refugees in Bangladesh in 2016 and 2017.

    The number of refugees from Myanmar in Bangladesh increased from 276,198 to 932,204 from 2016 to 2017. The increase in Bangladesh accounts for the overall increase in the number of refugees. Therefore, it is reasonable to believe that the centrality remains almost the same because the increase in refugees is mostly focused on a single country of residence.

    Next, you'll compare the Myanmar data with the Rwandan genocide, which took place against the Tutsi minority in 1994.

  17. On the Year filter, click 1994. On the Origin filter, deselect Myanmar and select Rwanda.
  18. On the map, zoom to Rwanda (in central Africa).

    Persons of concern originating in Rwanda in 1994

  19. On the Migration of persons of concern map, point to the node for Rwanda.

    The outdegree is 37. The Refugees, asylum seekers, and internally displaced persons table indicates there were more refugees than internally displaced persons. The countries with the most persons of concern from Rwanda were the Democratic Republic of the Congo, Tanzania, Burundi, and Uganda. The Total Population values for each country are also higher than the number of refugees by hundreds of thousands of people.

    Because there is no data in the Asylum-seekers column of the table, you'll change that column to show different data.

  20. On the Refugees, asylum seekers, and internally displaced persons table card, click the Asylum-seekers header and choose Returned refugees.

    Change the asylum seekers column to returned refugees

    The summary table is updated. The total number of returned refugees is over 1.2 million people, meaning that 1.2 million Rwandans had refugee status at some point in 1994 and then returned to Rwanda before the end of the year. The large number of returned refugees is different from what you saw in Myanmar, where the refugees, asylum seekers, and internally displaced persons accounted for most of the persons of concern.

    One difference between the Rwandan genocide and the other situations that you have looked at so far could be the length of the conflict. For instance, the genocide in Myanmar and the civil war in Syria are both multiyear conflicts beginning in 2017 and 2011, respectively, and continuing to the present day. In comparison, the Rwandan genocide took place over an approximately 100-day period from April 1994 to July 1994, meaning the movement of refugees from their home country and back could happen within the span of a single year.

  21. Use the Year filter to repeat the analysis for the years from 1992 to 1995.

    The following table summarizes the data for the specified years:

    YearOutdegree centralityRefugeesReturned refugeesInternally displaced personsTotal population

    1992

    11

    434,736

    140

    0

    434,876

    1993

    28

    450,462

    307

    0

    450,769

    1994

    37

    2,257,573

    1,208,005

    300,000

    3,765,578

    1995

    50

    1,819,366

    238,745

    0

    2,058,111

    Like the case in Myanmar, the genocide in Rwanda saw a drastic increase in the number of refugees in the year of the genocide, although Rwanda also had an increase in internally displaced persons and Myanmar did not. The data for Rwanda in 1994 and 1995 also shows the return of persons of concern to Rwanda after the end of the genocide and in the following year. Similar trends cannot be analyzed in Myanmar because the genocide has not yet ended.

    Another notable difference between Myanmar and Rwanda is the change in centrality. The outdegree in Myanmar remained almost the same from 2015 to 2017, whereas Rwanda saw an increase in outdegree each year, including the year after the genocide.

Compare changes in movement over time

Your final analysis will compare past and current refugee situations. Ethiopia and Afghanistan both had conflicts in the 1980s, 1990s, and 2000s that caused millions of people to become refugees. You'll look at how the movement of refugees from these countries has changed over time.

The Ethiopian Civil War took place from 1974 to 1991. The number of refugees from Ethiopia passed 1 million for the first time in 1979.

  1. On the Origin filter, deselect Rwanda. For Myanmar, click Edit filter.
    Edit filter for Myanmar

    The New Filter pane appears.

  2. Search for and deselect Myanmar. Search for and select Ethiopia.
  3. Click Apply.

    The Origin filter updates to show Ethiopia and Rwanda.

  4. Replace the Rwanda filter with Afghanistan.
  5. On the Year filter, choose 1979.

    Migration of persons of concern in 1979

    The overall patterns on the flow map are easier to see at this point in time than in later years because there are fewer connections between countries. There are also a large number of countries with nodes but no links, especially in western Europe and the Americas. The reason these countries are unconnected is because the countries do not have data for countries of origin. Instead, the origins are listed as Various/Unknown.

  6. On the Origin filter, check Ethiopia.

    Persons of concern originating in Ethiopia in 1979

    According to the map, the refugees from Ethiopia in 1979 were residing in eastern African and Middle Eastern countries. Although there could have been refugees from Ethiopia living in countries that did not collect data on country of origin, this map shows that Ethiopians in 1979 were seeking refuge in relatively few countries.

  7. Point to the node for Ethiopia.

    The outdegree is 12.

  8. Point to the links to Somalia, Sudan, and Saudi Arabia (the three thickest links).

    The largest number of persons of concern was 1,175,000 in Somalia.

    Next, you'll look at how the patterns changed over time. The number of refugees from Ethiopia reached a record high in 1980. The Ethiopian Civil War ended in 1991, making 1990 the final full year of conflict. You'll also look at the most recent data in 2017.

  9. Use the Year filter to look at the pattern of refugee movements for Ethiopia in 1980, 1990, and 2017.

    The movement in 1980 is mostly the same as 1979, with the exception of Italy, which is the first recorded European country to have refugees from Ethiopia. The outdegree in 1980 was 15.

    By 1990, the countries of residence have changed to include Nigeria, Sweden, the United Kingdom, Italy, Denmark, and the United States, although the outdegree of 14 shows that the countries with Ethiopian refugees has changed, rather than increased. According to the numbers on the Refugees, asylum seekers, and internally displaced persons table, the number of refugees in a single country has also decreased, with approximately 850,000 refugees in Sudan, compared to over 2 million refugees in Somalia in 1980.

    Finally, in 2017, the number of persons of concern from Ethiopia once again surpassed 1 million. However, this time the table shows that the majority were internally displaced persons within Ethiopia, and the refugees were spread out across most of the world, with an outdegree increasing to 104.

    Next, you'll look at Afghanistan. You'll start with 1980, the year after the Soviet-Afghan War began.

  10. Change the Year filter to 1980.
  11. On the Origin filter, deselect Ethiopia and select Afghanistan.

    Persons of concern originating in Afghanistan in 1980

  12. Use the nodes and Refugees, asylum seekers, and internally displaced persons summary table to find the outdegree and learn more about the persons of concern from Afghanistan.

    There were over 1.7 million refugees from Afghanistan in 1980, with over 1.4 million of them residing in Pakistan, and the outdegree shows that Afghan refugees were residing in only five countries. Like Ethiopia in the early 1980s, most of the Afghan refugees stayed within the Middle East, near their country of origin. Italy is the only country in Europe with known Afghan refugees. Once again, it's important to remember that many countries did not collect data on country of origin at the time.

    Next, you'll look at how the patterns in Afghanistan changed over time. The number of persons of concern from Afghanistan was at a record high in 1991. The War in Afghanistan began in late 2001 and continues to this day.

  13. Use the Year filter to look at the pattern of refugee movements for Afghanistan in 1991, 2002 (the first full year of the War in Afghanistan), and 2017, the same as you did for Ethiopia and Afghanistan in 1980.

    In 1991, the number of refugees from Afghanistan was over 6 million. Most of the refugees were residing in Pakistan and Iran, with the rest residing in several countries in the Middle East and Europe, China, and the United States. The outdegree was 12.

    In 2002, most refugees were still living in Pakistan and Iran, but the number of countries with refugees from Afghanistan has increased and is spread across all continents. This change could be explained by the increase in countries that report countries of origin in their data. The numbers of refugees in countries outside of the Middle East also increased. The outdegree increased to 81.

    In 2017, the number of internally displaced persons in Afghanistan was almost as large as the number of refugees. There were also more countries across Europe with a high number of refugees from Afghanistan, and more countries in Africa and South America with Afghan refugees. The outdegree increased to 100.

  14. Save your workbook.

    Two noticeable trends emerged by looking at the data over time: the increase in internally displaced persons and the increase in outdegree. The high levels of internally displaced persons is consistent with the analysis on Syria and Colombia as well. Therefore, you may want to focus some of your reporting on the current refugee crisis on the context and causes of internally displaced persons. The increases in outdegree were especially noticeable in Afghanistan, where the outdegree increased with each new year you analyzed. There are a few possible explanations for these changes; for instance, the data collection methodology improved over time, providing more accurate data on country of origin. Other global changes, such as increased use of air travel, could also explain why refugees and asylum seekers are residing in more countries over time.

In this lesson, you used Insights for ArcGIS to explore the global flow of refugees through space and time, fueled by war, genocide, and other destabilizing events. The exploratory analysis you completed can be used to spark further inquiries and analysis into refugee crises and provide a quantified visual to communicate with your audience.

Throughout the lesson, you became familiar with link analysis, which is an analysis technique that allows you to identify and analyze relationships within a dataset using a network of nodes and links. This technique can also be used for analysis in other industries, such as public health or crime analysis.

You can find more lessons in the Learn ArcGIS Lesson Gallery.