|
|
Post by therealriga on Sept 25, 2022 19:36:56 GMT
Well, so long as the 2023 review is still live, I hope Plan Builder will stick to showing the boundaries currently being used for that review. After that ... who knows?
Yes, the ward boundaries being used for the current review will remain the default until that review is complete. Next steps will be to add overlays for revised recommendations when they become available, and to update the census statistics part of the website once the 2021 ward level data is released. When I sent my WhatsApp Christmas list to Santa one of the requests on it was to have the wards and electorate figures from previous reviews (the ones done pre-1997 and 2010, for example.) I guess that's never going to happen due to the work involved though, right? |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Sept 26, 2022 11:18:14 GMT
Boundary Assistant still has the 2015 boundaries and electorates as an option, so I imagine it'd still have the Dec 2020 ones even if more recent ones were added. The most obvious reason to do that might be the upcoming Scottish Parliament review. Eh, wards are rather pointless for the scotish parliament reviews, the seats are too small for Scotlands very large wards to be any real use. Polling district's or lower are more appropriate for the task in my opinion. Are devolved assemblies and county councils going to be added to your site? |
|
haroldthepolitician
Labour
'You can be just what you want to be, Just as long as you don't try and compete with me' - Jarvis
Posts: 216 
|
Post by haroldthepolitician on Feb 15, 2023 15:03:33 GMT
Eh, wards are rather pointless for the scotish parliament reviews, the seats are too small for Scotlands very large wards to be any real use. Polling district's or lower are more appropriate for the task in my opinion. Are devolved assemblies and county councils going to be added to your site? I might look for some time to add functionality for scotland but i dont have the shapes for scotland sadly - the code for the site is a monstrosity and If i had the time i'd completely rip it up and rewrite it. |
|
|
|
Post by bjornhattan on Feb 15, 2023 15:20:33 GMT
Are devolved assemblies and county councils going to be added to your site? I might look for some time to add functionality for scotland but i dont have the shapes for scotland sadly - the code for the site is a monstrosity and If i had the time i'd completely rip it up and rewrite it. One issue is that the OA boundaries in Scotland are even more chaotic than the ones in England (due to a combination of much lower populations per OA and extremely low population densities in some areas). For instance, North Ballachulish is in an OA with the entire north coast of Loch Leven, a few houses on the fringes of Kinlochleven, and then a random salient which reaches almost all the way to Corrour station. If I were you, I'd use wards (or possibly communities) as a base building block with the ability to split these freely and population weighted centroids being used to compute the figures - but I don't know how viable this is from a web design point of view. |
|
haroldthepolitician
Labour
'You can be just what you want to be, Just as long as you don't try and compete with me' - Jarvis
Posts: 216
|
Post by haroldthepolitician on Feb 21, 2023 21:31:15 GMT
You'll be very happy to know, I plan to redevelop the website over the next few months. This will include a complete revamp of the UI - A rehashed map renderer; that includes ward/parish/town overlays and the [ability to split OA's (population weighted by postcodes)]*, more tools for mapping including a paintbrush and filler (allowing you to fill OA's (split if boundary crosses) included in a ward/parish/town/constituency) and hopefully the introduction of Scotland and Wales. I also hope the framework im using will make the website generally easier to use and more efficient. At a later date, I may add an account system and map storage.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Feb 22, 2023 12:22:50 GMT
Will you also be reforming how multi-member wards are dealt with? I still find that bit difficult to use
|
|
|
|
Post by jamesdoyle on Feb 22, 2023 16:16:07 GMT
haroldthepolitician: hi, I've only just seen this mentioned, so I've taken a look. I can't seem to get anything to work - I can select a council, but neither left- nor right-clicking on an OA seems to make anything happen - nothing changes colour, and there's no list of draft wards with sizes I can add to or take from. There's a long white rectangle on the left within nothing in it. I've tried in both Chrome and Edge. How do I get started? Cheers
|
|
Ports
Non-Aligned
Posts: 606
|
Post by Ports on Feb 22, 2023 17:09:27 GMT
haroldthepolitician : hi, I've only just seen this mentioned, so I've taken a look. I can't seem to get anything to work - I can select a council, but neither left- nor right-clicking on an OA seems to make anything happen - nothing changes colour, and there's no list of draft wards with sizes I can add to or take from. There's a long white rectangle on the left within nothing in it. I've tried in both Chrome and Edge. How do I get started? CheersNot Not Harold of course, but if it makes it any quicker, I believe that issue is because you need to enter a number of wards (specifically, the number of wards desired plus one) in the textbox for each council first. |
|
haroldthepolitician
Labour
'You can be just what you want to be, Just as long as you don't try and compete with me' - Jarvis
Posts: 216
|
Post by haroldthepolitician on Feb 22, 2023 18:56:39 GMT
haroldthepolitician : hi, I've only just seen this mentioned, so I've taken a look. I can't seem to get anything to work - I can select a council, but neither left- nor right-clicking on an OA seems to make anything happen - nothing changes colour, and there's no list of draft wards with sizes I can add to or take from. There's a long white rectangle on the left within nothing in it. I've tried in both Chrome and Edge. How do I get started? CheersNot Not Harold of course, but if it makes it any quicker, I believe that issue is because you need to enter a number of wards (specifically, the number of wards desired plus one) in the textbox for each council first. This^ |
|
haroldthepolitician
Labour
'You can be just what you want to be, Just as long as you don't try and compete with me' - Jarvis
Posts: 216
|
Post by haroldthepolitician on Feb 22, 2023 18:58:36 GMT
Will you also be reforming how multi-member wards are dealt with? I still find that bit difficult to use Yes, what I may do, is allow a user to input the amount of seats and then show the quota for (1) seat and then if you increase its size the seat at the end of the list is deleted (in the code) and added to the bigger seat. Otherwise, I'll just allow the seats in each ward to be defined prior to drawing. |
|
|
|
Post by manchesterman on Feb 23, 2023 12:12:46 GMT
This looks like a fantastic resource haroldthepolitician ! Would there be the possibility to show the electorate figure for each 'unit' when you hover over it [as with Boundary Assistant where the electorate for each ward is shown]? Unless I'm missing something the only way you can discover this figure is by checking the amended cumulative total for the chosen district after you've clicked/highlighted it. Other than that small tweak, I have to say this is a brilliant tool which looks like it will keep being improved upon!  |
|
haroldthepolitician
Labour
'You can be just what you want to be, Just as long as you don't try and compete with me' - Jarvis
Posts: 216
|
Post by haroldthepolitician on Feb 23, 2023 23:43:32 GMT
This looks like a fantastic resource haroldthepolitician ! Would there be the possibility to show the electorate figure for each 'unit' when you hover over it [as with Boundary Assistant where the electorate for each ward is shown]? Unless I'm missing something the only way you can discover this figure is by checking the amended cumulative total for the chosen district after you've clicked/highlighted it. Other than that small tweak, I have to say this is a brilliant tool which looks like it will keep being improved upon! Version 2 will have this ability + (demographic data will be shown too) . As i'm using a framework this time, this stuff is alot easier. Also going to implement a system to predict a rough political leaning for seats (I'll stop short on notional results because creating a reasonable model has, after consulting someone who taught R and statistics to 2nd year Uni students, told me is unreasonable at least using traditional linear regression models). |
|
|
|
Post by bjornhattan on Feb 24, 2023 0:32:30 GMT
This looks like a fantastic resource haroldthepolitician ! Would there be the possibility to show the electorate figure for each 'unit' when you hover over it [as with Boundary Assistant where the electorate for each ward is shown]? Unless I'm missing something the only way you can discover this figure is by checking the amended cumulative total for the chosen district after you've clicked/highlighted it. Other than that small tweak, I have to say this is a brilliant tool which looks like it will keep being improved upon! Version 2 will have this ability + (demographic data will be shown too) . As i'm using a framework this time, this stuff is alot easier. Also going to implement a system to predict a rough political leaning for seats (I'll stop short on notional results because creating a reasonable model has, after consulting someone who taught R and statistics to 2nd year Uni students, told me is unreasonable at least using traditional linear regression models). As a stats graduate, I'd gently dispute what you were told about creating a reasonable notional model - the hardest part there isn't actually the stats* but working out which variables to use in your model. If you can think of a dozen or so variables which are available at OA level, aren't strongly correlated, and predict political leanings, you're most of the way to having a reasonable demographic model. Though in fairness, because I enjoy this sort of thing I'm perhaps underestimating the difficulty and it may be harder to someone who doesn't like csv files and/or has a life. * There is one minor stats issue you'll run into - to be technical linear regression produces unbounded predictions, whereas vote shares have to be greater than 0% and less than 100%. The easiest way around this is to model a function of the vote share so that no matter what prediction you get it has a sensible interpretation - if you're familiar with "log odds" that's what I tend to use. Also, because vote shares have to add up to 1 you might find slightly strange results, so I usually modelled proportion of the electorate or population who voted for each party rather than proportion of voters! |
|
haroldthepolitician
Labour
'You can be just what you want to be, Just as long as you don't try and compete with me' - Jarvis
Posts: 216
|
Post by haroldthepolitician on Feb 24, 2023 11:53:10 GMT
Version 2 will have this ability + (demographic data will be shown too) . As i'm using a framework this time, this stuff is alot easier. Also going to implement a system to predict a rough political leaning for seats (I'll stop short on notional results because creating a reasonable model has, after consulting someone who taught R and statistics to 2nd year Uni students, told me is unreasonable at least using traditional linear regression models). As a stats graduate, I'd gently dispute what you were told about creating a reasonable notional model - the hardest part there isn't actually the stats* but working out which variables to use in your model. If you can think of a dozen or so variables which are available at OA level, aren't strongly correlated, and predict political leanings, you're most of the way to having a reasonable demographic model. Though in fairness, because I enjoy this sort of thing I'm perhaps underestimating the difficulty and it may be harder to someone who doesn't like csv files and/or has a life. * There is one minor stats issue you'll run into - to be technical linear regression produces unbounded predictions, whereas vote shares have to be greater than 0% and less than 100%. The easiest way around this is to model a function of the vote share so that no matter what prediction you get it has a sensible interpretation - if you're familiar with "log odds" that's what I tend to use. Also, because vote shares have to add up to 1 you might find slightly strange results, so I usually modelled proportion of the electorate or population who voted for each party rather than proportion of voters! That was my biggest issue, OA concentrate too many similiar demographic - so much so they go way past what i can reasonable interpolate. for example, theirs an OA in my authority which is 55% pensioners whereas the highest constituency in that regard is something like 15%. |
|
|
|
Post by bjornhattan on Feb 24, 2023 12:49:03 GMT
As a stats graduate, I'd gently dispute what you were told about creating a reasonable notional model - the hardest part there isn't actually the stats* but working out which variables to use in your model. If you can think of a dozen or so variables which are available at OA level, aren't strongly correlated, and predict political leanings, you're most of the way to having a reasonable demographic model. Though in fairness, because I enjoy this sort of thing I'm perhaps underestimating the difficulty and it may be harder to someone who doesn't like csv files and/or has a life. * There is one minor stats issue you'll run into - to be technical linear regression produces unbounded predictions, whereas vote shares have to be greater than 0% and less than 100%. The easiest way around this is to model a function of the vote share so that no matter what prediction you get it has a sensible interpretation - if you're familiar with "log odds" that's what I tend to use. Also, because vote shares have to add up to 1 you might find slightly strange results, so I usually modelled proportion of the electorate or population who voted for each party rather than proportion of voters! That was my biggest issue, OA concentrate too many similiar demographic - so much so they go way past what i can reasonable interpolate. for example, theirs an OA in my authority which is 55% pensioners whereas the highest constituency in that regard is something like 15%. I agree - though equally the strongest OA for the Conservative party nationally will likely have voted more than 80% Tory (if not 90%) but even Castle Point was only 76%. When I created by own notionals (which were at LSOA level), I used a random forest technique which essentially cut values off at the highest point in the constituency data set. Any LSOA which was more than that would be treated the same as one that was at that highest level. But because of this (and the specific model I used), I ended up with seats that were too uniform, so I had to adjust the LSOA figures afterwards to make them more varied within seats. Needless to say, I wouldn't recommend a similar approach! |
|
haroldthepolitician
Labour
'You can be just what you want to be, Just as long as you don't try and compete with me' - Jarvis
Posts: 216
|
Post by haroldthepolitician on Feb 24, 2023 18:54:07 GMT
That was my biggest issue, OA concentrate too many similiar demographic - so much so they go way past what i can reasonable interpolate. for example, theirs an OA in my authority which is 55% pensioners whereas the highest constituency in that regard is something like 15%. I agree - though equally the strongest OA for the Conservative party nationally will likely have voted more than 80% Tory (if not 90%) but even Castle Point was only 76%. When I created by own notionals (which were at LSOA level), I used a random forest technique which essentially cut values off at the highest point in the constituency data set. Any LSOA which was more than that would be treated the same as one that was at that highest level. But because of this (and the specific model I used), I ended up with seats that were too uniform, so I had to adjust the LSOA figures afterwards to make them more varied within seats. Needless to say, I wouldn't recommend a similar approach! An approach I may consider is using a polynomial regression instead in the hope those higher values show a tapering off on values - then we'd have to extrapolate which isnt the ideal solution. |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Feb 25, 2023 18:06:38 GMT
I've noticed a few of the Hertfordshire districts link to the wrong places
Clicking on Stevenage takes you to Bromsgrove
Clicking on St Albans takes you to St Edmundsbury
Clicking on East Herts takes you to Arun
EDIT: Clicking on Welwyn Hatfield takes you to Surrey Heath
|
|
haroldthepolitician
Labour
'You can be just what you want to be, Just as long as you don't try and compete with me' - Jarvis
Posts: 216
|
Post by haroldthepolitician on Feb 25, 2023 20:58:26 GMT
I've noticed a few of the Hertfordshire districts link to the wrong places Clicking on Stevenage takes you to Bromsgrove Clicking on St Albans takes you to St Edmundsbury Clicking on East Herts takes you to Arun EDIT: Clicking on Welwyn Hatfield takes you to Surrey Heath That'll be the welsh and scottish codes in my file messing things up - as im not really focused on repairing the old one you might just have cope for now. Fixing that might take a while. |
|
haroldthepolitician
Labour
'You can be just what you want to be, Just as long as you don't try and compete with me' - Jarvis
Posts: 216
|
Post by haroldthepolitician on Jun 26, 2023 13:39:56 GMT
Hey guys, Doing this project for my Computing NEA; would be lovely if i can receive some responses on this form! forms.gle/P2eUduZXvYUC5pS88 |
|
|
|
Post by carlton43 on Jun 26, 2023 14:25:03 GMT
Hey guys, Doing this project for my Computing NEA; would be lovely if i can receive some responses on this form! forms.gle/P2eUduZXvYUC5pS88Two observations about your form from a quick read. At the top : correct the spelling of questionnaire. In the last question on the first page delete 'alternative' as it is a misuse and is not needed. The very last question is muddled and needs to be worded in a simpler form. |
|
