Visualizing FIDE chess data (maps, graphs, plots)

Any thoughts?

That's really cool.

Thanks.

Very nice

Wow, good job. Very interesting graph for Slovenia; females have a kind of multimodal distribution, with a peak at ~1300 and another one at ~2000.

One interesting find I did on the side (that I will include in the future) was compare females of each country with the Overall Female population and the Overall Male population.

There were two particular major countries, for female players, that I found that had higher concentrations at higher levels (over 2000 ish) than the overall Male population. Those two countries were the U.S and Ukraine. I find that to be interesting.

Well, to be fair, the female population is pretty small there, so it's not surprising this variation may be present. All that the density curve does is smooths and normalizes (if you compare Sexes) out the histogram.

Oh, yes, you are right, of course. 

I have added an age distribution feature and a dynamic populating barplot graph (add more than one country to use the last relevant tab) to the app, the original link works again, but not for long. Hurry and use it before it expires!

https://infiniteflashchess.shinyapps.io/FIDE/

Georgia's graphs are very interesting, with the women's curve rising above the mens at fairy high ratings.

That one is interesting, yes. (app is down right now)

Here is a screenshot of what glamdring27 is talking about:

I briefly explored why this happened at the 2000 rating mark, by partitioning it by K-factor.

It turns out, that if you look at only chess players who only have K-factors of 20 in georgia, women perform incredibly well relative to men.

Here is how the number of Georgian players are partitioned by Sex and K-factor if you were interested.

It could be also that georgian males are relatively worse (in regards to concentration) than males overall. Also, the female size is of that K-factor is still small.

The two countries with the relatively higher echelon of female players, percentage wise, compared to female overall and males overall were Ukraine and the U.S.A

This is a very good application already, however you can make it better by indicating somehow the exact definition of terms that you are using.

What does the frequency of players within countries mean? Is this the absolute number of active, rated players? For an outsider this may not be clear.

It would be nice to have the same chart for other categories as well: all registered players, rated players, active rated players, inactive rated players, etc. In general there should be a main filter for this that is valid for all possible stats not just the country frequency chart.

Thanks for telling me that. I will include text boxes under every single graph, detailing what each graph represents.

Yeah, I'll add a dropdown bar for those player groups you mention.

How about filtering by main categories? ( unrated / rated , active / inactive , all age / middle age )

Edit: sorry, your first version of the post did not mention this.

Also add charts for activity rates, female participation rates by various categories ( per age, per country ).

I'm mostly interested in middle age ( 20 - 40 ) players. This is a very important category. For examlpe 27 of the 33 female unrestricted grandmasters are of middle age ( between 20 and 40 years old ). This shows them in a much more favorable comparison, though they are still underperforming their participation. There are practically no old female unrestricted grandmasters since for women to become unrestricted grandmasters was almost unheard of decades back. Practically all old gms are male which distorts stats.

Okay, 2nd to last visually cool thing I will do for this app. I couldn't help it. 

This is a globe of the of the size of the chess populations (done through the shinyGlobe package) for each country. The longer the line, the greater the number of players there are in a country.

I'll be back focusing on more statistically oriented things soon enough.

I don't see the point of including unrateds to be honest.

I now have active/inactive included in the app. I have had age group included since the beginning. You just can just type in the minimum and maximum ages into the box on the left side.

Unrated players have a significance. Any unrated player on the Fide list has something to do with chess, so it is an information about the interest in chess in some country or at some age. For example I'm on the Fide list because for a short period I played at Fide online arena. I have a Fide ID and even had a blitz rating but for some reason I'm counted as unrated.

As far as middle age is concerned a convenience filter that sets the age limits for you would be helpful. If I regularly browse this page, then it is inconvenient to set both the lower and upper limit of age always I make a query for some usual category. Some preset ranges would be handy like middle age, Fide categories like U8/ U10/ U12/ U14/ U16/ U18/ U20 etc. Users like common sense defaults.

I am not sure how to include unrated players actually. This is somewhat problematic to include because all of the graphs I have are responsive to the filters I have below (except the map and globe). One of these filters is that the person must be rated in the first place. I could create a filter that would look at the original dataset of 550000, and just divide it into rated, unrated, and both, this would only be useful for the barplot, map, global.

You think users would find using a slider bar (the blue bar with cirlces on the ends below) more of a pain than regular fixed inputs?

I think it wouldn't be such a big deal, but I can create some kind of dropdown menu with the list of age groups, sure. That will be done later I suppose.

I'm still thinking about how to go about the participation chart. It's only useful if I have yearly data. I really don't want to format all of the crappy formatted data FIDE has posted on their website. It may be time for me to check out your program you mentioned in the other thread....

On 2nd thought, isn't the participation rate chart my density curve I post in the app itself? I have it stated below.

This is just a smoothed curve of the points you plotted in the other thread.