maemo.org - Talk - MeeGo Coding Competition 2011

maemo.org - Talk (https://talk.maemo.org/index.php)

- Community (https://talk.maemo.org/forumdisplay.php?f=16)

- - MeeGo Coding Competition 2011 (https://talk.maemo.org/showthread.php?t=71561)

Re: MeeGo Coding Competition 2011

Quote:

Originally Posted by neal (Post 1131092)

??? My script generates a ranking.

I wanted to say the problem with frictional transfer, and not your script, was that it doesn't give ranking.
I'm trying to find a solution which would use the same algorithm only with ranking.

Re: MeeGo Coding Competition 2011

Quote:

Originally Posted by mmlado (Post 1131093)

diff? Sorry, no.
I did it only on Business & Office Category, by hand. So run election for 1 seat, copy the winner. Add winner to the withdrawn list, run the election again.
I think it should give same results as frictional transfer, only we would have ranking.

I still don't understand. My script generates a ranking. Why do you think it doesn't?

Re: MeeGo Coding Competition 2011

Quote:

Originally Posted by mmlado (Post 1131097)

I wanted to say the problem with frictional transfer, and not your script, was that it doesn't give ranking.
I'm trying to find a solution which would use the same algorithm only with ranking.

My script does that.

Re: MeeGo Coding Competition 2011

Quote:

Originally Posted by neal (Post 1131099)

My script does that.

Your script uses fractional transfer?
In that case I'm sorry. I've lost track. You have my support. :)

Re: MeeGo Coding Competition 2011

Quote:

Originally Posted by mmlado (Post 1131105)

Your script uses fractional transfer?
In that case I'm sorry. I've lost track. You have my support. :)

My script uses FTSTV (which is what was used in the last maemo community coucil election, as I understand it) and generates a ranking, yes.

Re: MeeGo Coding Competition 2011

I'm really sorry to ask... but if the "fractional transfer" from mmlado and neal does the same but in different ways... why are the results different?

Quote:

Originally Posted by neal (Post 1128987)

MeeGo Coding Competition Business & Office Category
1 MeePasswords http://competition.meetmeego.org/submissions/68
2 LUCID iron http://competition.meetmeego.org/submissions/4
3 ClipMan http://competition.meetmeego.org/submissions/13
4 MeeToDo http://competition.meetmeego.org/submissions/74
5 Wallet http://competition.meetmeego.org/submissions/18
6 NewsG http://competition.meetmeego.org/submissions/76
7 Quick Milk http://competition.meetmeego.org/submissions/78

Quote:

Originally Posted by mmlado (Post 1131085)

We already have a mechanism to remove candidates, by saying they have withdrawn. It's even easier to run my proposal.

Quote:

MeePasswords http://competition.meetmeego.org/submissions/68
ClipMan http://competition.meetmeego.org/submissions/13
Wallet http://competition.meetmeego.org/submissions/18
LUCID iron http://competition.meetmeego.org/submissions/4
MeeToDo http://competition.meetmeego.org/submissions/74

Here are the results I got, in less then 5 minutes :) :

So, where is the difference if I remove a person after the first run or if I say the person has withdrawn? :confused:
There must be a small mistake in one of those algorithm... or have I overseen something?

Re: MeeGo Coding Competition 2011

Both scripts generate a ranking, both are based on the STV used in the council election, but they are not necessarily equivalent (ie I cannot prove nor disprove it)

By construction, mmlado's method (which is the one I proposed some pages ago, but he has a working implementation) will give a consistent ranking.

Not sure about neals one (ie, can it produce {A}, {A,D}, {A,B,C} out of the A,B, C and D participants?) as far as I know it didn't happen, but I don't know if it can.

Re: MeeGo Coding Competition 2011

Quote:

Originally Posted by Helmuth (Post 1131126)

I'm really sorry to ask... but if the "fractional transfer" from mmlado and neal does the same but in different ways... why are the results different?

So, where is the difference if I remove a person after the first run or if I say the person has withdrawn? :confused:
There must be a small mistake in one of those algorithm... or have I overseen something?

Good point. It would be good if mmlado would indicate exactly how he came up with his results (preferably with a script so it is easy to reproduce and debug).

Thanks.

Re: MeeGo Coding Competition 2011

Quote:

Originally Posted by Helmuth (Post 1131126)

Good question. Maybe it's that by withdrawing, usually apps win by reaching the threshold and not by elimination.

As I see it we both get the apps in list that would be returned by FTSTV if we run them for the appropriate number of seats, only the ranking is different. And that can be a big difference for the winners.

Re: MeeGo Coding Competition 2011

Quote:

Originally Posted by Helmuth (Post 1131126)

So, where is the difference if I remove a person after the first run or if I say the person has withdrawn? :confused:

As far as I understand, neal's approach does not remove candidates. Instead the process is as follows:

Run election for one seat. This one will be the winner. (This step is common for neal's and mmlado's approach). In the example MeePasswords gets chosen and is selected as first.
Next: run election for two seats. Now MeePasswords and LUCID iron get elected. MeePasswords was already chosen as first place so LUCID iron is chosen for second place.
Next: run election for three seats. Now MeePasswords, LUCID iron, and ClipMan get elected. MeePasswords and LUCID iron had already been assigned to places 1 and 2 so ClipMan is chosen for place 3.
etc.

mmlado's approach explicitly eliminates candidates for the next voting run by saying already chosen candidates have withdrawn. I don't know the mechanics/mathematics well enough but the different results are most probably caused by the remaining votes being distributed differently among the remaining candidates as compared to neal's approach.