how to predict elections from rough polling data

[Roedy Green]

Election prediction.

To predict an election you really need riding by riding polls.
However, these are expensive, so the best you get is region by region
polls.

I wondered if it might be possible to take those polls, combine them
with historical polling data, historical voting data, and combine that
with a few selected riding by riding polls to more accurately predict
an election.

I was thinking of writing this up as a student project. I wonder what
sort of hints I could give the student on how to proceed. It may
require some sort of random simulations, fuzzy logic, or neural nets.

 

[Arne Vajhøj]

Election prediction.

To predict an election you really need riding by riding polls.
However, these are expensive, so the best you get is region by region
polls.

I wondered if it might be possible to take those polls, combine them
with historical polling data, historical voting data, and combine that
with a few selected riding by riding polls to more accurately predict
an election.

I was thinking of writing this up as a student project. I wonder what
sort of hints I could give the student on how to proceed. It may
require some sort of random simulations, fuzzy logic, or neural nets.

Wrong group again ?

Or is it intended to be coded in Java ?

Arne

 

[JK]

I wondered if it might be possible to take those polls, combine them
with historical polling data, historical voting data, and combine that
with a few selected riding by riding polls to more accurately predict
an election.

Nate Silver is already doing essentially the same thing, if I
understand you correctly : http://www.fivethirtyeight.com

(Yeah, I hate when that happens, too.)

-- JK