Proximity map of Europe

[gixajo]

Didn't try that, but I don't know if those are good labels (rich vs poor), its just a hunch feeling.

I've got the data from Kaggle.

Then I used scikit-learns standard scaler and k-means.

But yeah, I know what you want, you want to see where each country is on a scatter plot. I might make it later on if I figure out how Vahaduo works.

You can chose the labels or reference names you like, which I said was only an example.

[Brás Garcia de Mascarenhas]

Portugal

Density (P/Km²) - 111

Greece

Density (P/Km²) - 81

---

Portugal

Migrants (net) -6,000

Greece

Migrants (net) -16,000

---

Portugal

Under-five mortality rate - 3.7/per 1.000 live births

Greece

Under-five mortality rate - 3.8/per 1.000 live births

---

Portugal

GDP Per Capita - $24590.40

Greece

GDP Per Capita - $24024.20

---

Portugal

Literacy rate for 2018 was 96.14%

Greece

Literacy rate for 2018 was 97.94%

---

Portugal

Phones - 1,260.28 per 1,000 people (2007)

Greece

Phones - 1,120.56 per 1,000 people (2007)

---

Portugal

Arable land (% of land area) - 10.7%

Greece

Arable land (% of land area) - 16.6%

---

Portugal

Fertility Rate - 1.29

Greece

Fertility Rate - 1.30

---

Portugal

Crude death rate - 11

Greece

Crude death rate - 11

---

Portugal

GDP - 237.69 USD Billion

Greece

GDP - 209.85 USD Billion

---

Portugal

Unemployment rate - 5.6%

Greece

Unemployment rate - 18.30%


I am failing to understand how Portugal is not on the second group in the first map and Greece is.

[gixajo]

[B]Portugal


I am failing to understand how Portugal is not on the second group in the first map and Greece is.

Not only you, in the first map Portugal should be in blue colour, simply because could be near Greece, but slighty better.

And in the second one, Italy and Spain should be in the same group, whatever is it, but in the same group, because Italy has slighty better "indicators" in data than Spain.

But maybe data in the Pulstar sources were different.

[vbnetkhio]

Didn't try that, but I don't know if those are good labels (rich vs poor), its just a hunch feeling.

I've got the data from Kaggle.

Then I used scikit-learns standard scaler and k-means.

But yeah, I know what you want, you want to see where each country is on a scatter plot. I might make it later on if I figure out how Vahaduo works.

k-means sucks.
run it a again with that same data, the results will be different.

[pulstar]

...

I'm going to double check it, perhaps there was some NaN value for Portugal.

[pulstar]

k-means sucks.
run it a again with that same data, the results will be different.

That's why I used kmeans++

[Chris596]

Didn't notice that, but to be honest it fits better with Georgia/Cyprus/Armenia anyway. But I think it should be blue here

Actually I think the 3 groups version is probably the most accurate
Blue: Western Europe
Yellow: Western/Civilized but less fortunate economically
Red: Balkaniggas and Slavomongols

That's me B0$$

[pulstar]

Not only you, in the first map Portugal should be in blue colour, simply because could be near Greece, but slighty better.

And in the second one, Italy and Spain should be in the same group, whatever is it, but in the same group, because Italy has slighty better "indicators" in data than Spain.

But maybe data in the Pulstar sources were different.

I am failing to understand how Portugal is not on the second group in the first map and Greece is.

I noticed something regarding this. Some data has NaN (not a number) values meaning that the cells in CSV file were left blank for some countries. Now, I did check Portugal and it had all values, but some of the other countries didn't.
In order to solve this I previously (in my original calculation) used the median values from the columns to fill the NaN values for the countries where its missing (in my previous calculation), but now I tried using mean values. Portugal moves to second group (Western European one). I also added three Baltic countries, but with median method Portugal is still somehow closer to Eastern Europe on k=2.

If you want to download the CSV file and check it on your own, here's the link. I can also send you my jupyter notebook if you're interested, just let me know.