Figure Friday 2025 - week 28

Dash Python
,
22

Hi @feanor_92,

that was a lot of work, some great visuals :rocket: I’m missing the historical movie chloropleth, did you remove it?

I think your app would improve for users who do not know the data, if you add a little (more) context to some pages.
Example: if I go to the chloropleth page, I could be easily tricked into thinking “the more yellow or the higher the number” the better. That’s how it usually works and at least I do not read every word on a screen. Assuming we prefer less corruption, it’s exactly the other way around. The more yellow & higher rank, the more corruption thoughts.

You could add context by using something like this for the chloropleth, with other values and text:

colorbar=dict(
                title="CPI Score",
                tickvals=[2, cpi_average,  98],
                ticktext=["Highly Corrupt", f"Median {cpi_average}", "Very Clean"]  
                )

or in another way. I only know this (changing ticktext for the colorbar) because I used it for the first time this ff :sweat_smile: And after that removed the map, but it works very well.

My thoughts, and I should study PCA, I still do not understand what I’m looking at, but that’s my thing.

23

Very nice app, @feanor_92 .
I agree with @marieanne 's suggestion regarding the colorbar. I tend to view the colorbar and immediately assume that the higher (more yellow) the number the better.

I liked the historical trends line chart. Interesting to see 2024 as one of the lowest points since 2012.

Can you explain how to interpret the PCA chart?

24

This app is going to be amazing when it’s done, @Ester .
What happens in the choropleth map when you select a country from the dropdown?

25

I will show You tomorrow on pycafe.

P.S. My birthday coincidentally almost fell on the same day as FF’s, but mine is actually today, which is why the cake icon is showing. :grinning_face:

26

oh, now I see that cake icon :slight_smile:

Happy Birthday Eszter. Did you do anything special for your birthday? How did you celebrate?

27

Hi @feanor_92
Can you please share. if possible, some images of your app?. Unfortunately I can see it on Pycafe from my phone.

Thanks in Advance

28

Thank you @Adam! I was invited to a nice restaurant for lunch;)

29

Thank you so much for this feedback—I actually didn’t even notice this yesterday :pensive_face:
I’ve just pushed an update that adds contextual guidance on the map. Now, right above the map, users will see a note:

“※ Lighter colors indicate less perceived corruption (better); darker colors indicate higher corruption (worse).”

I’ve also customized the colorbar ticks to clearly label “Highly Corrupt,” “Median CPI,” and “Very Clean.” I hope this makes the dashboard more intuitive for first‑time viewers :slightly_smiling_face:

30

Thanks for the kind words! :folded_hands: I’ve also updated the colorbar per @marieanne suggestion so that the tick labels (“Highly Corrupt,” “Median CPI,” “Very Clean”) make it clear that lighter shades are better and darker shades worse.
As for PCA tab:

  • PCA1 (x‑axis): the combination of all numeric CPI measures that explains the most variation among countries.
  • PCA2 (y‑axis): the next most important, independent direction of variation.

In short, the PCA chart is a quick way to spot groupings or unusual cases beyond just the single CPI score: Countries that close together have very similar overall CPI profiles.

31

Hi! @Avacsiglo21!
Yes, of course :slightly_smiling_face:
Add pictures:
Снимок экрана 2025-07-17 151614
Снимок экрана 2025-07-17 154057
Снимок экрана 2025-07-17 155327
Снимок экрана 2025-07-17 151647
Снимок экрана 2025-07-17 151654

32

@feanor_92
Are you free to join us this Figure Friday session to go over your graphs?

33

HI, This topic is very interesting. I find it notable how the range of CPI values can be visualized, and with that purpose, I have developed two complementary charts. Both aim to highlight clustering patterns and the temporal evolution of perceived corruption levels at the country level.

cap_1
cap_11195×547 70.1 KB

cap_2
cap_21195×547 52.8 KB

Application code

34

Excellent thank you very much

35

Really cool PCA scatter graph. I really like this type of analysis, well done. One question the clústers you identified or use k means clustering?.

36

Interesting the Spyral chart, interesting analysis

37

Dissmiss the question, I re-read you use quartile for grouping :flexed_biceps:

38

@adamschroeder Thank you for invite!
Yes, I’d love to join the session. And I’ll try not to spread my post-cold cough to anyone :sweat_smile:

39

Thank you so much! :folded_hands:
In this version I didn’t run a formal clustering algorithm—those “groups” you see are simply countries colored by their CPI‐quartile (High → Very Low corruption). The PCA scatter just projects the data, and any visual clusters emerge naturally from those quartile labels.