The PAP prevailed accurately based on their model !

[serenditpity]

http://sammyboy.com/showthread.php?92009-PAP-believes-it-will-win-with-60-vote-share

Remember the thread and a couple of other threads initiated by me ?

The PAP won with the results of the model (60.1%) and all my predictions came true.

Opposition will get 1 GRC (aljunied) and 1 SMC (Hougang)
PP will be lost.

 

[ahsoo]

Yes your insider info is indeed very accurate. Hope you can share more often in future.

 

[lockeliberal]

Dear seren

Dammm that modeling is accurate :-)). It even took into account party prefernces and needs. JC escaped their notice and I really expected one more but tough.

My own estimations were to WP specific. The swing was against the government but also party, personality and persistence specific. lesson learned and noted.



Locke




QUOTE=serenditpity;724400]http://sammyboy.com/showthread.php?92009-PAP-believes-it-will-win-with-60-vote-share

Remember the thread and a couple of other threads initiated by me ?

The PAP won with the results of the model (60.1%) and all my predictions came true.

Opposition will get 1 GRC (aljunied) and 1 SMC (Hougang)
PP will be lost.[/QUOTE]

 

[ivebert]

 

[serenditpity]

Hi Locke:

Since we have known each other for a long time through this forum, let me say this with all humility:

I had a big role to play in devising the 2006 model and also tweaking the model between March and april this year. The model is not without it's flaws and limitations and based on the data set that I last saw on May 3rd, I felt good about the assumptions that I made. The model cant really cater to some of the smaller SMCs as efficiently as to the GRC and the larger SMC.

If the heavy weights of WP had also devoted some time to JC and of Chiam had reached out to some of the voters in PP, both could have gone to opposition. My model predicted that the differences in both constituencies would be about 5.5%, which got validated. The model also depended on the data that was available in 2006.

Dear seren

Dammm that modeling is accurate :-)). It even took into account party prefernces and needs. JC escaped their notice and I really expected one more but tough.

My own estimations were to WP specific. The swing was against the government but also party, personality and persistence specific. lesson learned and noted.



Locke




QUOTE=serenditpity;724400]http://sammyboy.com/showthread.php?92009-PAP-believes-it-will-win-with-60-vote-share

Remember the thread and a couple of other threads initiated by me ?

The PAP won with the results of the model (60.1%) and all my predictions came true.

Opposition will get 1 GRC (aljunied) and 1 SMC (Hougang)
PP will be lost.

[/QUOTE]

 

[Charlie99]

Hi Locke:

Since we have known each other for a long time through this forum, let me say this with all humility:

I had a big role to play in devising the 2006 model and also tweaking the model between March and april this year. The model is not without it's flaws and limitations and based on the data set that I last saw on May 3rd, I felt good about the assumptions that I made. The model cant really cater to some of the smaller SMCs as efficiently as to the GRC and the larger SMC.

If the heavy weights of WP had also devoted some time to JC and of Chiam had reached out to some of the voters in PP, both could have gone to opposition. My model predicted that the differences in both constituencies would be about 5.5%, which got validated. The model also depended on the data that was available in 2006.

**********

Hi Brother [serenditpity],

Now I understand how and why you were so accurate.
Congratulations.
I agreed with your suggestions re: WP vis-a-vis Joo Chiat, and SPP vis-a-vis Potong Pasir.
The leaders of WP and SPP should have appeared as a guest speaker at the respective rallies at Joo Chiat and Potong Pasir.

 

[lockeliberal]

Dear Seren

Your modelling takes into account the voter preferences for opposition parties . Impressive.



Locke

 

[travelbug]

Serendipity, can I ask you whether your model input has Meet the People sessions data on type of feedback, problems, resolutions exchanged between constituents & MPs. And added on to this, you have MPs walkabouts to gather feedback being incorporated into the model. My wildest guess is that MPS session data is the bedrock of this model, correct?

Also, a smaller base of voters in each constituency will tend towards some inaccurate results as opposed to bigger GRCs with larger voters. What I mean here is recall & precision inverse relationship.

 

[kukubird58]

hahaha....bro serend, nice model and good results.....
but like i said before it is basically more of good guesstimate because there are not much data
from GE2006 to start with.....many new voters in GE 2011.

If one follows GE2011 closely, with or without the model, one roughly know:

WP likely to win HG and Aljuneid GRC.
PP - a close fight which can go either way
among other reasons, papee won in the end becos:
1. new voters from new condos
2. Sitoh is established in PP after spending 10yrs in PP
Papee overall percentage around 60%.

The rest is really a bridge too far... although JC' close fight was a pleasant surprise.
Unless there is a revolution, it is unrealistic to expect wild swings in 5 yrs.
however the trend is good for Opp and Singaporeans overall......75.3% (GE2001), 66.6% (GE2006) 60.1%(GE2011).
it will forced Papee to take necessary actions to stop the rots....
the coming 5 years will be crucial to determine whether we are really on track towards a 2 party system.....Papee and WP.

In a nutshell....all opps supporters should close rank and support 1 strong opp party (WP in this case) to be really effective.
Really the rest of the opps parties were fodders for papee and pulled down the overall opps percentage.

 

[nato33]

**********

Hi Brother [serenditpity],

Now I understand how and why you were so accurate.
Congratulations.
I agreed with your suggestions re: WP vis-a-vis Joo Chiat, and SPP vis-a-vis Potong Pasir.
The leaders of WP and SPP should have appeared as a guest speaker at the respective rallies at Joo Chiat and Potong Pasir.

Joo Chiat SMC is too small to have its own rally. The WP candidate for Joo Chiat (JJ Yee) spoke at the Bedok Stadium rally for the East Coast GRC and at least another 2 WP rallies IIRC. In fact, at all the WP rallies, all candidates including the leaders appeared as a team but up to 10 spoke due to time factors

 

[Bigfuck]

Whatever the stupid model, if the votes go up another 5%, you may suddenly see a lot more seats as there is an overflow of opposition votes that cannot be contained in one losing constituency.

 

[serenditpity]

U will appreciate if I dont give all the details, wont U?

>>My wildest guess is that MPS session data is the bedrock of this model, correct?<<

Important, but this is not the bedrock.

>>Also, a smaller base of voters in each constituency will tend towards some inaccurate results as opposed to bigger GRCs with larger voters. What I mean here is recall & precision inverse relationship. <<

Yes, this is correct

Serendipity, can I ask you whether your model input has Meet the People sessions data on type of feedback, problems, resolutions exchanged between constituents & MPs. And added on to this, you have MPs walkabouts to gather feedback being incorporated into the model. My wildest guess is that MPS session data is the bedrock of this model, correct?

Also, a smaller base of voters in each constituency will tend towards some inaccurate results as opposed to bigger GRCs with larger voters. What I mean here is recall & precision inverse relationship.

 

[serenditpity]

U have to give it to me that the model predicted the vote share around 60%+/- share nation wide and it was that accurate.

I was also accurate in predicting the precise 2 seats, though in this forum(see all the threads) and outside kopitiam talk, everyone was predicting more GRC wins.

So, just give this to me lah

hahaha....bro serend, nice model and good results.....
but like i said before it is basically more of good guesstimate because there are not much data
from GE2006 to start with.....many new voters in GE 2011.

If one follows GE2011 closely, with or without the model, one roughly know:

WP likely to win HG and Aljuneid GRC.
PP - a close fight which can go either way
among other reasons, papee won in the end becos:
1. new voters from new condos
2. Sitoh is established in PP after spending 10yrs in PP
Papee overall percentage around 60%.

The rest is really a bridge too far... although JC' close fight was a pleasant surprise.
Unless there is a revolution, it is unrealistic to expect wild swings in 5 yrs.
however the trend is good for Opp and Singaporeans overall......75.3% (GE2001), 66.6% (GE2006) 60.1%(GE2011).
it will forced Papee to take necessary actions to stop the rots....
the coming 5 years will be crucial to determine whether we are really on track towards a 2 party system.....Papee and WP.

In a nutshell....all opps supporters should close rank and support 1 strong opp party (WP in this case) to be really effective.
Really the rest of the opps parties were fodders for papee and pulled down the overall opps percentage.

 

[cheekenpie]

Is she still laughing?

 

[kukubird58]

U have to give it to me that the model predicted the vote share around 60%+/- share nation wide and it was that accurate.

I was also accurate in predicting the precise 2 seats, though in this forum(see all the threads) and outside kopitiam talk, everyone was predicting more GRC wins.

So, just give this to me lah

hahaha......i have nothing against your model.......you predicted correctly......that's a fact....
so as many other bros who predicted correctly without using any models.
anyway JC and PP could easily have thrown a spanner in your prediction.....

 

[cheekenpie]

U have to give it to me that the model predicted the vote share around 60%+/- share nation wide and it was that accurate.

I was also accurate in predicting the precise 2 seats, though in this forum(see all the threads) and outside kopitiam talk, everyone was predicting more GRC wins.

So, just give this to me lah

Bro, in the other thread you said your source told you about the model. So how is it you help develop it? *confused

 

[TITW]

Prediction model: