Data Mining Pokemon Spawns in Singapore

data-mining-pokemon-singapore

So every blog/company is jumping on the Pokemon bandwagon to “unveil Pokemon nests around Singapore”. Here’s one more, but backed with data.

I’ll be revealing what data has revealed about the pattern of Pokemon spawns in Singapore in this post. In another post, I will discuss about the data collection, dataset and techniques used. I’ll try to leave technical findings and discussions to the other post and leave the juicy ‘actionable’ findings here.

Update: New Pokemon nests has been uploaded and can you find them in this post.

TLDR Version

I’ve plot mentioned nests on this map. You will find additional data on the map like count and standard deviation (in metres). 68% of the data falls within one standard deviation if that matters for you.

Legend

For each Pokemon, I’ll be performing a clustering algorithm on it. On top of the clusters, you will be able to see the number of spawns in that cluster (color of circle) and the spread of pokemon spawns (size of circle). A small red circle show the gold pot.

AeroAerodactyl – 142

Total Count: 47

Out of 700k data points, only 47 is Aerodactyl. You have better chance at winning 4D than walking into an Aerodactyl. However, it seemed like the spawn location for Aerodactyl is pretty predictable.

There are two major spawn nest, one at City Hall (1.292, 103.849) and another at Mapletree Business City (1.275, 103.799). From the cluster diagram you see a small red circle, showing the spawn spot but from raw data, I’ve uncovered an exact coordinate City Hall (1.29238660433637, 103.849206760772) that has spawned 17 of them!

Other Notable Locations (probability of spawns here are very low):

  • Botanic Garden (1.311, 103.815)
  • Clementi (1.326, 103.778)
  • Yio Chu Kang (1.395, 103.845)

Now run to City Hall if you want a Aerodactyl for yourself!

SnorSnorlax – 143

Total Count: 59

Having caught 4 Snorlax around my area in Jurong, I thought my area will be a hot spot, but nope. There seemed to be an even spread of Snorlax around Singapore.

There is a red spot at Tampines (1.349, 103.938) with 8 Snorlax. LIKE OMFG 8 SNORLAX IN LITTLE LESS THAN 2 DAYS?!? Other than that, there are a few more spots with moderate occurrence of Snorlax. Two honourable mentions will be NEX (1.348, 103.875) and Chong Pang (1.431, 103.831).

Other Notable Locations:

  • Seng Kang (1.39, 103.903)
  • Boat Quay (1.29, 103.848)
  • Holland Village (1.315, 103.79)
  • Ngee Ann Poly (1.336, 103.774)
  • Choa Chu Kang (1.38, 103.749)

Tampines here I come!

*Coordinates on the google map are updated locations

LaprasLapras – 131

Total Count: 30

Out of the two Lapras I own, none of them were caught in the wild. Looking at the numbers you might know why. It’s one helluva exclusive creature.

It seemed that the Pokemon has a fixed set of spawn spots. Most of the occurrence are on Sentosa (1.257, 103.822). The rare Pokemon can also be found at Gardens by the Bay (1.283, 103.861) and East Coast (1.306, 103.933).

If you’re thinking about catching a Lapras at the long kang below your house, continue dreaming.

DragDragonite – 149

Total Count: 19

Between breeding my own Dragonite and catching a wild one, I prefer the prior. A wild Dragonite may not have the best IV values and is also extremely rare. However, for those who wants a wild Dragonite, here you go!

The best location to catch a Dragonite is at Esplanade area (1.298, 103.859) and (1.293, 103.843)! You can see a couple of high density, high count spots on the cluster map.

Other Notable Locations:

  • Pasir Panjang (1.273, 103.798)
  • (1.396, 103.762)
  • Bukit Batok (1.352, 103.769)

PoryPorygon – 137

Total Count: 29

Have you met anyone with a Porygon? Nope? That’s because it is very rare as well. However, because it is not a battling Pokemon, no one talks about it. Have you thought about how it feels? No.

This blob of computer program seemed to spawning around Ang Mo Kio (1.365, 103.857), Bugis (1.303, 103.856), Marine Parade (1.305, 103.903) and Bukit Timah (1.326, 103.801). There are also occasional spawns in random location in the southern parts of Singapore.

ChanChansey – 113

Total Count: 20

Have you seen a Chansey? Me neither.

Contrary to popular beliefs, Chansey does not appear near hospital. Now stop bothering the sick people there, unless you are there to drop a lure.

From the data, it seemed like there are no fixed spawn points for Chansey. There are two points of interest that may potentially be spawn points, but the results are very inconclusive.

Points of Interest:

  • Promenade (1.294, 103.860)
  • Sentosa (1.256, 103.820)

Afterthoughts

While these findings are backed by data, using past data to extrapolate future spawn location may not be the best idea. As a result, I do not recommend going to mentioned place just because it was marked as a potential spot. Well, unless its a Aerodactyl or a Lapras…

I’ll be working on nests of more common Pokemon like Pikachu, Growlithe, etc on a future post. Let me know if you have particular interests in any Pokemon.

All findings will be plotted on this map.


 

Want to know more about data science?

Also, if you are interested in learning data science, want your kids a head start or want an interesting way to teach data science, fill in this form.


 

Disclaimer

There are currently 800k data points in the dataset. These data points are collected over the span of last 48+ hours. While the spawn data are 100% accurate, I’ve not made direct connection with the game servers. Data collected are from a third-party source. The raw data include the following information: latitude, longitude, pokemon, despawn time.

 

4 Comments

  • Rainald Lim
    September 1, 2016 at 9:25 am

    Errr porygon? There spawned 4 porygons at vivo yesterday. 2 of them were within an hour apart.

    • root (Author)
      September 1, 2016 at 10:14 am

      Yep the data for Porygon is from 4 days ago and does not include the data from yesterday. Thanks for the heads up

  • Zav
    September 5, 2016 at 2:52 pm

    2 days might not be representative of the spawns since weekdays and weekends might differ.
    Can perhaps get data for a week at least?

    • root (Author)
      September 5, 2016 at 4:16 pm

      Thanks for your suggestion. You do raise an important point. The results are inconclusive until it is validated. The post serves as a preliminary analysis. I’ll be doing cross-validation of the data when I find some time.

Comments are closed.

Hey You!

Want information on data mining Pokemon Go?
Or on the upcoming bitcoin introduction session?

Thank You!