The Pretty Charts thread


#1

Calling all Pinsters! I’m looking for anyone willing to contribute their very useful data, currently floating around the Pin in miscellaneous forms. I’m playing with some ideas for prettifying it. You would still “own” the data and get to update it whenever you want, according to your own schedule.

To kick the tyres of it, I started with metalmike’s PPR numbers from here. (h/t to metalmike for the collaboration so far). The object was to capture the monthly and quarterly numbers, year-on-year changes, and so on, all in one compact chart instead of the “wall of text”.

Here is the result of the initial tyre-kicking: note the numbers are not real (yet). Click on the screenshot below to go to the real interactive version in a new window/tab. I need to discuss further with metalmike before this is finalised.

i.imgur.com/ATzCe5g.png

The charts are in beautifully resizable SVG – no more fugly distorted bitmaps. The underlying technology is the wondrous D3.js library. The actual data is in simple csv or tsv form like this (I was hoping it could also be edited in Google sheets, but this hasn’t worked out [yet]):

Xlabel,2013,2014,2015,2016 Jan, 400, 603, 1254, 125 Feb, 400, 603, 1254, 125 Mar, 1100, 1200, 1300, 0 Apr, 1255, 1807, 1042, 0 May, 1161, 1149, 1035, 0 Jun, 1272, 2277, 1764, 0 Jul, 1255, 1807, 1042, 0 Aug, 1161, 1149, 1035, 0 Sep, 1272, 2277, 1764, 0 Oct, 1255, 1807, 1042, 0 Nov, 1161, 1149, 1035, 0 Dec, 1272, 2277, 1764, 0

Volunteer data contributors would be able to keep the data updated without having to worry about how the chart is done, while hard-core hackers can contribute their hand-coded charts. Here’s the code for the bar chart above. Bar charts are only the beginning. D3 allows data to be attached to any SVG images. Here’s one borrowed from Wikimedia Commons and tidied up with Inkscape:

Once in the right form it’s a (relatively) simple matter to tag the counties with data, and create heat maps, chloropleths etc. I’m hoping to do this with my own price/sq.ft. data.

Ideally we could keep some kind of structured list of pretty chart links, highlighting which ones had been updated recently. Anyone interested in prettifying their existing data or charts? Doesn’t matter whether you are planning to hack it yourself, or just contribute data and let someone else design the visuals. Reply here or PM me.


#2

are you able to grab the gps location data from daft for houses along with the price drop data and price itself.
Be interested to interpolate the price drop data and the average price data. Classified by apartment/house/site/number of bedrooms/ sq footage/ or a combination of each.

If you want someone to do that I can certainly help :slight_smile: I`ve also done some work with open layers.


#3

Interesting idea. For now, though, I’m not looking to do new data sources or scrapes, just to collate and prettify data that people are already producing on the Pin on a regular basis. Indeed, some people are already producing their own pretty charts. This is just an idea-starter for those that aren’t. For new data, I’d say go and produce/scrape it first, then think about how it might be presented visually.


#4

Very interesting.


#5

Gonna have a stab at creating an Ireland map example today. Code will be updated every now and again at:

PinsterDev on GitHub Gist

I’ll keep a mini-project diary there for the people who said they were interested in hacking.
bl.ocks.org is a cheap place to host our running code. You can see it here:

Map example on bl.ocks.org

When something is running there’ll be a link there to view in separate window mode (without the code/supporting files).


#6

Fair play PS, great work


#7

Eureka! A snapshot of asking prices in the first quarter of 2011 … I’ll spruce this up and stick up the singing, dancing, animated version in a day or two.


#8

Ok, here is version 0.1 … I need to fill in values and tooltips on the maps plus various tidy up. You can see values and YoY changes on the table view for now. The animation of the Ireland map gives you an idea of how things have centred on Dublin since 2012, even if only with colours. You should be seeing something like this when you click the link … I have only tested on Firefox so far. In of a bit of a rush, so wanted to get a version out there. Comments, suggestions and bug reports welcome.

i.imgur.com/snxFdqh.png?1


#9

How’s the raw data generated - the price per square foot/county and post code?

I’ve done a bit of work on mapping stuff using the R and the shape files from the CSO
this on population statistics for instance:

I’d love to play around a bit with property price data, but depends on the data!


#10

Great work, very interesting.

It doesn’t work in Chrome. :slight_smile:


#11

Or Safari on iPad, at least not on the train :smiley:
Will get to it shortly.


#12

It’s webscraped quarterly from myhome.ie.

Here’s the data and the current code.

gist.github.com/pinsterdev/898435e2f8561941b628


#13

Not sure if I should be relieved, but it seems to break in the same way on Chrome, Safari and Edge. I presume the table bit worked? – it does for me in all three. Hopefully the maps are just a small scaling problem (famous last words). Will look at that, plus adding text to maps, tomorrow.


#14

Thanks.

I think I’ve found a chunk of code that allows me to pull geo-coordinates from google using an address.
If I can get it to work I can run the PPR against that, then use the CSO maps to plot the data based on either small areas or electoral areas.
If I do it for Dublin it might give a clearer indication of what’s happening there.
Such a pity there isn’t decent size information in the PPR.

There goes today’s productivity.


#15

That sounds neat. Interested to see the results. I tried using Google APIs a while back to get coordinates from PPR addresses but free usage was limited to a few hundred translations per day, and I ran out of patience. Hope you don’t hit the same limits.


#16

I’ve gotten to 2617 on the 2015 list for Dublin (it’s 15k long). It’s kicked me out now saying I’m over my limit.
It’s enough to start writing the rest of the code with though I probably won’t get to it til later.


#17

Yeah, I think I reckoned it was going to take two or three months to do the whole PPR :neutral_face:


#18

How about using the ECAD?

A single-user license is only €60. Happy to chip in.

eircode.ie/docs/default-sou … f?sfvrsn=2


#19

Two probs as I see it … would cost €3k to use on a website, and you would have to write a parser/fixer-upper for the shitty PPR addresses, which Google already had. I tried this before and decided it was too much work.


#20

v0.2 now up. Now works on Firefox and Chrome. Should be at least viewable in Safari but may be some sizing bugs. (Would appreciate if someone would check it out on Safari on Mac to see if it differs from iPad). Doesn’t work on MS Edge or IE. Map text and tooltips are next. Comments, suggestions and bug reports welcome.

Click the image.
i.imgur.com/3JKspDF.png?1