This post is part 1 of the "Building a Data-Grounded Country Comparison Service" series:
The idea of comparing countries to choose the best one for living, working, or traveling comes to many digital nomads and remote workers. And there are already services that help with that, like the famous Numbeo. I personally love using it. However, it is crowdsourced, which means its data points are user-submitted and can be biased or outdated. Being data/ML-addicted, I wanted to create a more data-grounded approach to country comparison. That’s where I came up with this service idea (which, again, is not completely new). However, I find it a great playground for experimenting with data collection, analysis, visualization techniques, service design, development, deployment, and more. Moreover, it’s only the starting point, and I plan to document the entire journey in this series of posts. It will be a real-time experience of building something from scratch, so I expect a lot of reflections, learnings, and maybe even failures along the way. Let’s see how it goes.
Something about country comparison and data analysis
The Idea
Given a set of criteria that matter to you, get a ranked list of countries that best match your preferences for living, working, or traveling.
That’s pretty much the core of the idea. However, it already raises several questions:
- Criteria can differ widely from person to person.
- The importance of each criterion can vary significantly; it’s a matter of personal preference.
- To get a ranked list, we need to gather data for all countries and criteria, which can be extremely challenging in some cases.
Let’s try to address these questions one by one.
Criteria
When thinking about criteria for comparing countries, several aspects may immediately come to mind. Here are several examples:
- “I want to live in a warm country by the ocean.”
- “I want to work in a country with a low cost of living and good internet connectivity.”
- “I want to feel safe and secure wherever I go.”
- “I’m OK with some level of underdevelopment if nature is beautiful and people are friendly.”
- “I should be able to easily learn the local language and culture.”
- “The legalisation process for remote workers should be fast and straightforward.”
- and so on…
So far, I narrowed down the criteria to the following categories:
| # | Criterion | What it means to me |
|---|---|---|
| 1 | Climate and Environment | I want to live in a sunny, warm place with clean air. It shouldn’t be too hot, though. I also prefer mild winters (I’m not a fan of snow at all), and I don’t like it when it’s rainy all the time. |
| 2 | Generous People ️ | I want to be surrounded by friendly and helpful people. I want to feel welcomed and accepted. |
| 3 | Safety and Security | I want to feel safe walking the streets late at night. The chances of being a victim of crime should be low. |
| 4 | Society Development Level | I prefer countries with clean, well-maintained streets and no urban decay, reliable transportation and social infrastructure (fast internet, dependable banking, good food), and accessible, high-quality education and healthcare. I also value a developed, diverse cultural scene with attractions and entertainment, and active scientific and research activity across society. |
| 5 | Economy Development Level | I want to live in a country with a stable and developing economy, a low unemployment rate, and good job opportunities. I also prefer countries with a reasonable cost of living and affordable housing. |
| 6 | Freedom, Rights, and Peace | I want to live in a democratic country with high levels of personal freedom, human rights, and political stability. I also value countries that are peaceful and have low levels of conflict and violence. |
| 7 | Language and Culture | I want to live in a country where I can easily learn the local language and culture. |
| 8 | Remote Work Legalisation | I want to live in a country that offers remote work visas or digital nomad programs with straightforward application processes, reasonable fees, and favourable tax policies for remote workers. It should be easy to obtain and renew the residence permit, and potentially citizenship later on. |
| 9 | Nature and Outdoors | I want to live in a country with beautiful nature, diverse landscapes, and plenty of opportunities for outdoor activities like hiking, biking, and water sports. |
| 10 | Proximity to Home | I want to live in a country that is not too far from my home country, so I can easily visit my family and friends. I also prefer countries with good flight connections and affordable travel options. |
That’s a good starting point. However, it already seems quite impossible to find a country that would score highly on all these criteria. That’s why we need to consider personal preferences as well.
Importance of Criteria
I will use the most straightforward approach here: let the user assign weights to each criterion based on its importance. The total country score can then be calculated as a weighted sum of individual criterion scores:
Each \(score(Criterion, Country)\) will be a value between \(0\) and \(1\) showing how well the country performs on that criterion. That way the final country score will also be between \(0\) and \(1\), where \(1\) means the country is perfect on all criteria.
Data Collection
That’s going to be the most challenging part. For each criterion, we will gather a set of metrics or indicators that can be used to evaluate how well a country performs on that criterion. For example, for the “Climate and Environment” we’ll have metrics like average summer/winter temperature, average annual rainfall, air quality index, Environmental Performance Index, etc.
The problem, however, is that not all metrics are available for all countries. Indicators like GDP per capita, Human Development Index, or Internet Penetration Rate are widely available, while others like Rudest Nations are available only for a handful of countries. I will try to use indicators that are available for most countries. When a dataset is missing data for some countries, I will try to find alternative sources and, if necessary, use imputation techniques to fill in the gaps. For instance, average summer temperature can be estimated based on the country’s latitude, elevation, proximity to the ocean, and temperatures of neighboring countries. That’s where we’ll use some Machine Learning :)
Caveats
The whole idea has several caveats that I’d like to acknowledge upfront:
- The criteria and their importance are highly subjective. I defined a set of criteria based on my personal preferences; however, the service will allow users to customize their priorities by adjusting weights to their own needs.
- The data sources may not be reliable or up-to-date. I will try to use reputable sources, but some datasets may still be outdated or biased.
- Imputation techniques may introduce errors or biases. When filling in missing data, the estimates may not be accurate and could affect the final country scores.
- Finally, even within a single country, there can be significant regional differences. A country may score highly on safety overall, but certain cities or regions may have higher crime rates. You just can’t assign a single score to a whole country like the US, China, or Russia. But that’s a price I’m willing to pay for the sake of simplicity!
Next Steps
I’ll start with the crucial part of the whole service: data collection and analysis. In the next few posts, I will define metrics for each criterion, find data sources, gather the data, and come up with imputation and transformation techniques to make the data usable for country scoring.