These are chat archives for FreeCodeCamp/DataScience

8th
May 2016
Brahma Reddy Chilakala
@bradd123
May 08 2016 07:08
@profoundhub Yeah, I want to get a job in a data science startup. What is your opinion of that nanodegree?
evaristoc
@evaristoc
May 08 2016 09:43

@krisgesling just a quick add to my comment just above. You might think it contradicts what I advised at May 03 12:04 :point_up:. For your first project I had the impression that you wanted to use colours to represent the differences between several different variables at the same time. Blue for countries with the largest amount of males, yellow for intermediate, etc. I might have it wrong.

Still, the fact is that the colour variation make more sense specially if what you want to represent is a gradient. A nice example is your most recent map: a colour gradient representing ranges of number of participants.

What your map is probably not showing is your story. If what you want to represent is the amount of minorities that participated in the survey, I would suggest that as the main colouring aspect of your map. The rest of the information (example how many respondents per country) is possibly marginal.

In my opinion, your story would be more interesting if you not only tell about the proportion of minorities claiming to filling in the questionnaire: it would be great if you can add information about their coding aspirations, resources they use, or perhaps salary expectations.

My personal opinion is that highlighting minorities by excluding majorities is as unfair as the opposite: I would suggest to prepare maps for female, male, or non-(fe)-male alike. An impartial view allows for discussion and observations.

Hope this helps!

@krisgesling additionally: the pie in the tooltip is not saying anything, I am afraid. Better to get rid of it, if you cannot work a better way to represent it... IMO
Kris Gesling
@krisgesling
May 08 2016 09:56
@evaristoc @SamAI-Software have updated the map with suggestions made, and added a % bar for those who identify as an ethnic minority in their country - not sure it really works that well but figured it was worth a try.
As you say I previously had the map coloured as a ratio non-men: men however I think it misrepresents the data as countries who only have 2 campers can look like they're leading a progressive revolution. I don't think there's enough data for most countries to make the claims it would be trying to make, even if it looks like a more interesting story.
Ultimately I was working to address the question of gender breakdown per country for integration into a broader set of visualisations. I haven't thought of a nice way to do that using country fill so for me the pie graph or a stacked bar graph is the best solution.
Sam Aiken
@SamAI-Software
May 08 2016 10:09
@krisgesling well done! Looks great! :+1:
However, there is still too much text in tooltip, so it's hard to digest
  • How about putting non-male&non-female into one word? May be abbreviation?
  • Also there is no need for "no response"
  • And just "ethnic minority" will look easier
Kris Gesling
@krisgesling
May 08 2016 10:18
I hid the 'no response' where it was 0 but I find those stats interesting still. I could do the same for Trans*, genderqueer and agender but the lack of people choosing any of those options is also interesting. Maybe both will get cut if FCC use it but for me, I like them. Good point on the ethnic minority label. I still need to optimise the whole thing for mobile devices too...
evaristoc
@evaristoc
May 08 2016 11:13

@krisgesling if you don't want to use the "Others" label then try an acronym and in some corner of the map try a legend explaining what means?
Your final work looks nicer! I don't still feel that the pie is adding info though...

I can tell you what I see:

  1. The usual proportion is 80-85% males over the rest, particularly Asian-European countries
  2. That proportion seemed higher in South-America, Africa and some Asian countries; but lower in North American countries, China and some Oceanic countries (Australia, Philippines,...).
  3. Ethiopie is an interesting case...
  4. There were about xxx amount of countries where there was at least 1 person affirming to belong to non-male/female gender categories.
  5. If we exclude countries when no-one reported to belong to non-male/female gender categories, you can notice that that segment reached 1-2% of the total respondents, with some exceptional cases reaching 4-5%.
  6. The more respondents, the more likely that the data was better.

Thinking... suppose that you have some data pre-digested as above. Could that work to guide user's information discovery even though it is a bit prescriptive? Let's say as a user I select to check the Affirmation 2: the map could highlight all countries that show over 80% and those which don't. Those countries that are exceptional should be highlighted apart...
The Affirmation 4 or 5 can also allows for highlighting (or opacity...). Etc.

@krisgesling I am just giving you ideas, please be free to select your way according to your interests and possibilities. No pressure! :)

Obs: a legend should be introduced at some point for sure...

And the borders between countries... I hope you can find a solution...

evaristoc
@evaristoc
May 08 2016 14:22
@Evaderei: How did you solve the issue with the twitter posts?
Daniel
@profoundhub
May 08 2016 16:24
@bradd123 go for it, it should give you a good foundation and community of students and educators to lean on!
Victor
@Evaderei
May 08 2016 18:57
@evaristoc I haven't
evaristoc
@evaristoc
May 08 2016 21:10
Send me DM later with the code? I will try to find time to check it and discuss it with you... @Evaderei