Jump to content

Talk:2013 Australian Senate election

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Auto generated tables

[edit]

I have run out preliminary tables for the Senate votes. This was quite structurally complex as I had to combine several data sources (AEC aggregates all unendorsed/ungrouped candidates, ABC did not publish group swings). There are a few issues I am aware of, so these aren't perfect:

  • Some of the party names differ between states in the AEC data, so sometimes the name was blank for the Coalition or the Greens (which I fixed if I saw them)
  • I haven't output the turnout, formal and informal totals, I will enter these manually shortly
  • For some reason, the Katter candidates did not output, I suspect it is the apostrophe in the party name causing an issue
  • Something went wrong with the way I did the ballot paper ordering, so some of them came out wrong, I fixed them where I spotted them but these could be a bit out
  • The Ungrouped candidates have not appeared—I may have to delve into the preference distribution lists to check these numbers as I am unsure how AEC and ABC have handled them
  • There's a line break at the end of each candidate list which doesn't need to be there but I couldn't be bothered fixing and it doesn't appear on the page
  • I've probably missed some wikilinks or am pointing to disambiguation pages in some cases

--Canley (talk) 23:48, 30 November 2013 (UTC)[reply]

Another issue: I used the ABC figures (aware that they are a model of the count only) as the AEC downloads aggregate the Ungrouped/Unendorsed votes and I needed the breakdown—however there are some discrepancies in the ABC vote figures and percentages and they don't seem to include the Ungrouped candidates at all, so I'll need to check all these against the AEC figures. I can get the detail from the AEC preference distribution so I may reimport and try again. --Canley (talk) 01:28, 1 December 2013 (UTC)[reply]
OK, I think it's all done. I've double checked each state table against the AEC data and made sure that the number of votes add up to the number of formal votes. The swings were from the AEC data so they should be OK. The big problem was using the ABC data (which I did because the AEC data downloads aggregate all Unendorsed (Group XX) and Ungrouped candidates into a single total). It appears that some of the minor groups/parties and ungrouped candidates with less than a certain number of votes were omitted on the ABC calculator, which threw out some of the percentages. Some of the vote numbers in the ABC data were one or two votes out, and so the sum did not match the formal vote totals. Western Australia was a complete nightmare, because of the recount I presume. I know the ABC Senate "results" are just a model, but I'm surprised they weren't updated when the votes were declared and the preference distributions were released, I wonder if they ever will be? Apart from these problems, this is a very structurally complex table which combines at least three different data sources—to do it without a database/script/program would take a very long time and would likely have many more manual transcription errors (speaking from experience of making such errors). It worked pretty well, I just need to be more careful about the source data, and standardise the party names between states so they match the wikilink name. --Canley (talk) 11:58, 1 December 2013 (UTC)[reply]
This is why I used to do them manually :P Timeshift (talk) 21:52, 1 December 2013 (UTC)[reply]
You're welcome to do them manually if you don't think there is any benefit to automating it, or that automating causes more problems/errors than it solves. I did a lot of checking and corrections here but I would have had to have done that whether they were manual or not. If I had used the ABC data manually there would have been the same errors (granted I would have spotted the 0.01 point discrepancies between the percentage and the swing for new parties).
As an example, I used the automated output of the HoR results to replace the ones in the Division articles which were done manually by some editors, but contained dozens of mis-spelt names, parties omitted, digits transposed, tables copied from other divisions where the candidate or party name was not changed, and almost every single turnout percentage was wrong. The current style of these Senate results tables was done by Frickeg two years after the 2010 election—if it were not for trying to duplicate this style with the candidate listings (as you originally had in the Senate results) I would have done a straight dump or transcription of the AEC data into a wikitable and not bothered with any script or automation (which, by the way, is what I did for the HoR by state summaries, where I made several manual errors). --Canley (talk) 22:54, 1 December 2013 (UTC)[reply]
No im good thanks :) Timeshift (talk) 23:08, 1 December 2013 (UTC)[reply]

Table style

[edit]

There's a discussion on table style here. Timeshift (talk) 23:44, 21 February 2014 (UTC)[reply]