exploring UNLI Neighbourhoods

A short guide

When you go to a new city you want to explore different neighbourhoods to get a feel for the city. Maybe you’re moving to a new country or maybe you’re just visiting a new city for a few days.

You can use UNLI Neighbourhoods to get a very high-level overview of different neighbourhoods in a city. You might want to see what the most vibrant or boring neighbourhoods are. The most safe or environmentally dangerous neighbourhoods. Or you might just want to see where the most people are at.

In this example I searched Singapore and sorted by the “Risk & Safety” column. Some risk is okay if the safety level is high. Definitely check other sources too because most of my data is spatial and so if a region doesn’t have spatial data then it could underestimate the risk or safety of the area. In this example “Little India” is a safe place because there is a lot of infrastructure. If you get hurt there are lots of people or things nearby to help like hospitals or police stations so don’t worry too much.

If we click the “Risk & Safety” column again then it will sort in reverse. At the bottom of the list Panbil Mall might be a little bit dangerous but it still has some stuff nearby that could help you. I include criminal activity, protest deaths, typhoon deaths, environmental factors like chemical plants in the risk column so it doesn’t necessarily mean that there is a lot of gang activity in that area. It is just a weighted number that represents a certain amount of risk calculated from the data that is publicly available.

But wait! I thought we were looking at Singapore? Why are there neighborhoods from other countries??

Yes. I realize this. Part of the reason is that it’s just easier to externalize/outsource my categorization of what counts as a city. I use the Natural Earth Populated Places dataset because it is a very careful balance between too many cities and too little. If I only used the top 500 biggest cities to group all the neighbourhoods in the world then there would be places thousands of kilometers away which are included in the nearest population center. My goal is to balance accuracy and exploration. There are arguably 300,000+ human settlements in the world but Natural Earth counts a little over 7,000. Far more than you could reasonably visit in one lifetime but it isn’t complete by any means.

In my example city of Singapore only 64 of the listed neighbourhoods are actually in Singapore. 27 Malaysian neighbourhoods are actually closer to Singapore than they are to the center of Johor Bahru or whatever else is the nearest Malaysian population center. 243(!) Indonesian neighbourhoods are closer to Singapore than they are to a population center that is included in Natural Earth’s dataset.

So why include half of Batam Island? It could be over a million people…

Well… my main reason is that I don’t have a better way of knowing what should be considered a population center vs what shouldn’t be. If I include too many “cities” then there will be no room for exploration. And that leads me to my second reason. I want people to explore and discover new neighbourhoods near their city (or the city they are visiting). The ferry from Singapore takes about 40 minutes to Batam Island while from Johor Bahru is around 90 minutes. If someone doesn’t ever have a reason to lookup Batam then they will never think of going to the neighbourhoods there. But there is a balance so we just group our 2million+ neighbourhoods by the nearest major population center.

Here is a voronoi diagram to show the area of influence that Singapore has:

So all the neighbourhoods in this region will be included as part of Singapore and some Singaporean neighbourhoods will be included into Johor Bahru—actually this is a little strange because I just looked it up and I didn’t see ANY Singaporean neighbourhoods in Johor. hmm… where is Yishun? where is Khatib? I will have to investigate further after I eat some cheese naan

here is a global map of coverage: https://i.postimg.cc/Pry49jZs/ts.jpg