The world’s leading publication for data science, AI, and ML professionals.

Using data science for social good: Migrants and their fatal routes

Data exploration of the migrants and their ultimate journeys

The past – a train journey

On a fine summer day in 2016, I was on a train from Zagreb to Vienna, travelling between the European capitals, admiring the scenery from the window, enjoying my small cup of coffee, and working on a client presentation. There was a brief scheduled stop at Varazdin, a small city around 50 miles away from Zagreb, where hundreds of people started pouring into the train; they were everywhere – sitting on the floor, near the toilets carrying with them bags that contained a small sample of life they once lived in their home countries. A family (I assumed) entered my compartment as well and I lifted my jacket that was lying on an empty seat; The woman passenger who was already in the compartment and I looked at each other and we had a transient non-verbal communication about the situation through our eyes and reluctantly accepted the things as they are.

Photo by Katie Moum on Unsplash
Photo by Katie Moum on Unsplash

After the train pulled from the station, the family seemed a little comfortable and took stock of all of their possessions. I started talking to them and asked where they are heading and in their broken English they said wherever it is safe and they can start a life. I learned that they were from Syria, both husband and wife were teachers in a village and reached the coast of Greece in a boat, and that’s when the conversation was interrupted by a small child who complained of hunger; the mother quickly conjured small quantity of oatmeal in a box and fished out baklava-looking long bread from a tin can, split it into 8 pieces and distributed among everyone in the compartment. I politely declined a few times but she was persistent and I accepted reluctantly. They stayed on the train for an hour or so and the rest of the journey was more or less silent with intermittent dialogue. We said bye to each other and I told them to stay safe and wished them all the luck for the rest of the journey. That was the end of that small interaction.

I was pensive and decided to write a blog post on this incident but I never did.

The present – an analysis

A few weeks ago, I stumbled upon the Missing Migrants Project and the associated data and it refreshed my memory of the train incident. So, I spent some time wrangling the data to find a few eye-opening facts.

This article isn’t a display of data science capabilities so I’ll refrain from adding code blocks to the article, but in case you want to follow along then the python code and data is present on GitHub. Also, I’ve decided in favor of more monochromatic visualisations given the subject matter. This is my humble attempt to throw some light on the fatal and perilous journeys a few have to take to save their lives.

The Missing Migrants Project is trying to capture as much info as possible about migrants and it comes from myriad sources – a few of them are legit and rest of them might be underreporting for various purposes. There is a lot of missing data esp. for…

  1. Number of survivors, 85% missing – People could have gone missing after disembarkment or political reasons such as to placate the citizens that not too many migrants are entering the country.
  2. Number of children who survived, 86% missing – Minor trafficking.
  3. Migrant routes, 55% missing – Due to policies survivors might refrain from telling the exact route they took or they might not even know where they were.

I treated the missing values and took a deeper look at the data. The data timeline is from Jan 2014 to Feb 2021.

Most prominent cause of death

Drowning causes most deaths (Image by author)
Drowning causes most deaths (Image by author)

As expected, drowning causes the most deaths. As I have gathered hundreds of people travel for weeks in small dinghies at the mercy of the wind to reach the shore. These numbers are estimated on the number of bodies recovered. You add vast seas and unforgiving stormy weathers to the picture and do the math. In my opinion, the numbers on the x-axis are lower.

Deadly years

Image by author
Image by author

As expected, 2016 was the year when most of the news on the migrant crisis was aired. 2020 sees a dip but it could be due to COVID-19 or lack of resources.

Deadly regions

The Mediterranean and northern Africa remain hotbed (Image by author)
The Mediterranean and northern Africa remain hotbed (Image by author)

The Mediterranean and northern Africa see the most deaths and that could be because of the sheer number of people trying to flee from the region. Two regions that I picked from the graph were the US-Mexico border and Southeast Asia. I do know about the Central American migration but I never equated it to the level of African/Middle East migration to Europe despite knowing about the authoritarian government of Honduras, Guatemala, and Nicaragua and failing economy of Venezuela. Also, Bangladeshi immigration to India and the Rohingya crisis in Myanmar are well-documented news but I didn’t have a number associated with them.

Let me check the scale of migration by adding the number of deaths and survivals.

Image by author
Image by author

Look at the sheer scale of attempts made at the Mediterranean front that data tells us. Out of ~60k attempts, 40k were successful which is greater than 65% success rate.

Cause of deaths

Given that most routes are water-based, I am expecting most deaths are due to boat capsizing and shipwrecks.

The Mediterranean data goes up to 10k, I limited it to 3k so one can see the heights of bars for other regions as well (Image by author)
The Mediterranean data goes up to 10k, I limited it to 3k so one can see the heights of bars for other regions as well (Image by author)

US-Mexico border is famous for shootings but somehow that isn’t getting captured here; it says the cause of death is Unknown(skeletal remains). Most people walk for months on feet and it is possible that their friends and relatives bury them wherever they find a place and that’s how skeletal remains come to the top.

Drowning remains the top cause of death for most migrant journeys.

Timeline of deaths

Sorry for the bad x-axis, I know that it can be made cleaner but I can't recall how. (Image by author)
Sorry for the bad x-axis, I know that it can be made cleaner but I can’t recall how. (Image by author)

News in the summer of 2016 was brimming with the European migrant crisis and the same is reflected in the above graph, with May 2016 hitting a peak.

The peak could be because of two reasons:

  • A 125-mile anti-migrant trench that runs along the Tunisian-Libyan border was completed in Feb 2016 (source)
  • Acute dry conditions in North Africa (source)

Summer months are expected to have more activity.

May, June, September see most deaths(image by author)
May, June, September see most deaths(image by author)

Hotbeds of migratory journeys

As seen from the data above, I am expecting Northern Africa – the region near Libya, Central America, and South-East Asia to be highlighted.

Plotting the same on a map reveals the truth which is in line with expectation.

Image by author
Image by author

Out of ~40k deaths in past 6 years, we don’t have the info to classify ~28k of them. It shows that all these numbers are estimated and the real number could be much much higher than what we have in front of us.

Image by author
Image by author

The future – an unknown abyss or hope?

Despite the worldwide attention, the crisis remains of monstrous proportions. A mass effort is going on to solve this problem but the dents made until now are largely insignificant. I understand that it is a multidimensional crisis and not at all an easy one to solve but one has to find a solution otherwise the numbers will keep on increasing.

I don’t know what happened to the family from the train; I hope that they reached where they wanted to and started a new life away from the turbulence.

Kein mensch ist illegal (No one is illegal)

No one is illegal (Source)
No one is illegal (Source)

Related Articles