Aaron Strife
Banned
Yeah but like .5% of voters in Pennsylvania switched from GOP to Dem omg Trump
-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
PoliGAF 2016 |OT3| You know what they say about big Michigans - big Florida
- Thread starter Spoiled Milk
- Start date
- Status
Suikoguy
I whinny my fervor lowly, for his length is not as great as those of the Hylian war stallions
I've seen a couple of people say that Trump post was some sort of Science joke.
The problem with that is, he has presented himself as someone who explains science to non-scientists. Why try to make extremely nerdy scientisty political themed jokes on twitter?
brainchild
Banned
Because he's actually a statistician who has actually explained his methodology himself, unlike the armchair analysts on these internet forums.
And saying his model is based on Google searches is an egregious affront to the work that he's put into his model.
Tyler uses Google TRENDS (which is based on searches, locations, time frames, etc.) to look at the demographical makeup of each state, and analyze specific variables that shift relative to one another, compared to how they shifted in previous results, and use them as a reference point in order to make predictions.
His model is based on math, inductive reasoning, and deductive reasoning, not 'Google searches and whatnot'.
benicillin
Banned
Tyson is the definition of "never meet your heroes." He's kind of an ass and I really don't enjoy his personality.
The point is that it isn't a very good model and he's been wrong more than he's been right. The only reason anyone is talking about him is that he managed to luck out and get Michigan right and even then he was way off on the margin.
EDIT: It's also rather telling he's not applied this same model to the GOP.
titiklabingapat
Member
This is getting real ugly real fast.
brainchild
Banned
Could be, which could prove to be a problem in the future.
Lol, ok
And yet, he has still been less wrong with each prediction (his most recent predictions didn't have as many states to predict).
brainchild
Banned
It might not be very reliable RIGHT NOW, but that may not be the case by, say, the beginning of April.
I don't see anything outrageous here except some ignorance of biology, which he then acknowledged in his next couple of tweets
It's lurking variables! Polling directly measures what we want. It turns out AlphaGo can't accurately predict the primaries.
Yes, but how reliable would it be in a situation with more than 2 people running? Like say what's been going on with the GOP? That's a completely different situation demographically and would his data points even be predictive there?
If he can make it work, great. I'll add him to the list, but as of right now I see no reason to bother with him.
Ogrekiller
Member
By the end of June his model will be 100% be able to predict the Democratic nominee.
Jack Remington
Banned
Doubtful. The guy just seems like kind of a joke as a modeler tbh.
By November he will be able to predict earthquakes
We finally have the correct model
Code:
5 + 3x + 9.2x^2 + 1.01x^3 + x^4 + 76x^5 + 540.1x^9 + 99x^17 + 88.5x^38 + 74.0x^409 + 99x^1001 + 0.0000000000001x^1100 + x^9000 + 35x^2.54e98Suikoguy
I whinny my fervor lowly, for his length is not as great as those of the Hylian war stallions
I would not take it that far. It's a valid area of research that is unproven.
User 463088
Banned
Gotten two calls from Hillary volunteers today. 
Makes the heart happy.
MSNBCs bumper for Tuesday is hilarious.
Will Trump hang on? Will Florida pull through for Rubio? Can Cruz beat expectations? Will Kasich secure his home state!?
And things will happen on the Democratic side too.....
Makes the heart happy.MSNBCs bumper for Tuesday is hilarious.
Will Trump hang on? Will Florida pull through for Rubio? Can Cruz beat expectations? Will Kasich secure his home state!?
And things will happen on the Democratic side too.....
Bam Bam Baklava
Member
One scenario that no one has covered is what if pollsters overreact to the result in Michigan and adjust their methodology too far in the other direction?
She is soooo damn lucky this year. In soooo many ways.Gotten two calls from Hillary volunteers today.
MSNBCs bumper for Tuesday is hilarious.
Will Trump hang on? Will Florida pull through for Rubio? Can Cruz beat expectations? Will Kasich secure his home state!?
And things will happen on the Democratic side too.....
Gotten two calls from Hillary volunteers today.
MSNBCs bumper for Tuesday is hilarious.
Will Trump hang on? Will Florida pull through for Rubio? Can Cruz beat expectations? Will Kasich secure his home state!?
And things will happen on the Democratic side too.....
Isn't having Clinton volunteers call each other kind of inefficient? This is the kind of waste president trump will cut from our federal government.
brainchild
Banned
Polling based on statistical inferencing has its limitations as well, such as relying on relatively old data to predict the composition of the electorate in future primaries.
I don't know, and I'm not arguing that you should bother with him, but I think we should all keep and eye out and see just how accurate his predictions can get.
Thanks for your opinion.
Certainly possible.
TumblingHippo
Member
Had a close friend today tell me he's voting for Trump if Hillary is the nominee, because "even if he's racist he'll do a good job". Must be nice being a white male and not worrying about the consequences of the Trump nomination.
I hate politics when they reveal things about "friends" like that...
I hate politics when they reveal things about "friends" like that...
It doesn't even count as overfitting if your model fails for many existing data points! That's just called being wrong.
So who does he plan on voting right now? Bernie?
This is why I think Bernie is bad for the country.
If Trump is elected we'll probably have Bernie to thank more than anyone.
Politics is a strange place where it's an exceedingly complex subject matter with a large learning curve to get into, yet people often think they can comment on it despite having no knowledge of the subject.
I try to educate people best I can when I hear something extremely off base, slip in a quick fact here or there. Let them puzzle out the correct conclusion.
My point is it doesn't matter how predictive he can get if it's only good for a 1-on-1 in the Democratic primary where one candidate has massive youth support. Unless he can expand it to cover the GOP or a larger number of candidates this will never be anything more than a curiosity.
theprodigy
Member
You know, you might as well post how his predictions have fared vs actual results:
2/27
South Carolina: prediction-Clinton 60.50%, actual-Clinton 73.88% (13.38% bias in favor of Sanders)
3/1
Minnesota: prediction-Clinton 52.02%, actual-Sanders 61.69% (13.71% bias in favor of Clinton, got the winner wrong)
Colorado: prediction-Sanders 52.94%, actual-Sanders 59.36% (6.42% bias in favor of Clinton)
Virginia: prediction-Clinton 68.49%, actual-Clinton 64.63% (3.86% bias in favor of Clinton)
Arkansas: prediction-Clinton 65.34%, actual-Clinton 69.03% (3.69% bias in favor of Sanders)
Massachusetts: prediction-Clinton 58.02%, actual-Clinton 50.72% (7.3% bias in favor of Clinton)
Oklahoma: prediction-Sanders 53.49%, actual-Sanders 55.54% (2.05% bias in favor of Clinton)
Alabama: prediction-Clinton 72.86%, actual-Clinton 80.22% (7.36% bias in favor of Sanders)
Texas: prediction-Clinton 59.42%, actual-Clinton 66.30% (6.88% bias in favor of Sanders)
Tennessee: prediction-Clinton 67.26%, actual-Clinton 67.09% (.17% bias in favor of Clinton)
Georgia: prediction-Clinton 73.99%, actual-Clinton 71.70% (2.29% bias in favor of Clinton)
Vermont: prediction-Sanders 71.33%, actual-Sanders 86.34% (15.01% bias in favor of Clinton, but I think this can be handwaved as a home state advantage)
3/5-3/6
Kansas: prediction-Sanders 57.85%, actual-Sanders 67.75% (9.9% bias in favor of Clinton)
Louisiana: prediction-Clinton 66.18%, actual-Clinton 75.42% (9.24% bias in favor of Sanders)
Nebraska: prediction-Sanders 62.28%, actual-Sanders 57.15% (5.13% bias in favor of Sanders)
Maine: prediction-Sanders 61.62%, actual-Sanders 64.39% (2.77% bias in favor of Clinton)
3/8 aka the day people actually started paying attention to this guy
Michigan: prediction-Sanders 53.48%, actual-Sanders 50.77% (2.71% bias in favor of Sanders)
Mississippi: prediction-Clinton 81.94%, actual-Clinton 83.39% (1.45% bias in favor of Sanders)
general thoughts: idk lol
(note that as far as margins of victory go, multiply the "bias" by 2, I also removed all non-Clinton/Sanders votes from the actual results to bring it in line with the predictions)
2/27
South Carolina: prediction-Clinton 60.50%, actual-Clinton 73.88% (13.38% bias in favor of Sanders)
3/1
Minnesota: prediction-Clinton 52.02%, actual-Sanders 61.69% (13.71% bias in favor of Clinton, got the winner wrong)
Colorado: prediction-Sanders 52.94%, actual-Sanders 59.36% (6.42% bias in favor of Clinton)
Virginia: prediction-Clinton 68.49%, actual-Clinton 64.63% (3.86% bias in favor of Clinton)
Arkansas: prediction-Clinton 65.34%, actual-Clinton 69.03% (3.69% bias in favor of Sanders)
Massachusetts: prediction-Clinton 58.02%, actual-Clinton 50.72% (7.3% bias in favor of Clinton)
Oklahoma: prediction-Sanders 53.49%, actual-Sanders 55.54% (2.05% bias in favor of Clinton)
Alabama: prediction-Clinton 72.86%, actual-Clinton 80.22% (7.36% bias in favor of Sanders)
Texas: prediction-Clinton 59.42%, actual-Clinton 66.30% (6.88% bias in favor of Sanders)
Tennessee: prediction-Clinton 67.26%, actual-Clinton 67.09% (.17% bias in favor of Clinton)
Georgia: prediction-Clinton 73.99%, actual-Clinton 71.70% (2.29% bias in favor of Clinton)
Vermont: prediction-Sanders 71.33%, actual-Sanders 86.34% (15.01% bias in favor of Clinton, but I think this can be handwaved as a home state advantage)
3/5-3/6
Kansas: prediction-Sanders 57.85%, actual-Sanders 67.75% (9.9% bias in favor of Clinton)
Louisiana: prediction-Clinton 66.18%, actual-Clinton 75.42% (9.24% bias in favor of Sanders)
Nebraska: prediction-Sanders 62.28%, actual-Sanders 57.15% (5.13% bias in favor of Sanders)
Maine: prediction-Sanders 61.62%, actual-Sanders 64.39% (2.77% bias in favor of Clinton)
3/8 aka the day people actually started paying attention to this guy
Michigan: prediction-Sanders 53.48%, actual-Sanders 50.77% (2.71% bias in favor of Sanders)
Mississippi: prediction-Clinton 81.94%, actual-Clinton 83.39% (1.45% bias in favor of Sanders)
general thoughts: idk lol
(note that as far as margins of victory go, multiply the "bias" by 2, I also removed all non-Clinton/Sanders votes from the actual results to bring it in line with the predictions)
brainchild
Banned
For you maybe, but I'm sure every media outlet in the world would be interested in a statistician who could [hypothetically] predict the results of any future Democratic primary with 99% accuracy, even if it's just 1 on 1.
blackaroni
Member
There's a town hall tonight AND tomorrow?!?!?!?!?
FloydtheFathead
Member
Superfreak.. Super freak.. He's superfreaky now..
Well that was basically how I would imagine that ass looking like.
Voysa Reezun
Banned
I have said this before, but besides the embracing of racism, sexism, homophobia, etc., the Republicans have done a terrible job of laying out the case for free trade beyond "growth!"
It turns out that their voter base finally asked "for whom?" in response to that argument.
I also think that their cleaving to Reagan's philosophy that growth should be distributed mostly to the top, and that social programs to address this are bad and bound to fail, in combination with their support of free trade deals, was always bound to end up with populists trying to take the controls of the party away from the Buckleyites, whose economic philosophy is intellectually bankrupt from the get-go because it benefits too few people.
People laughed at Reagan's "most scary words are 'I'm from the government, and I'm here to help," joke, but it has always been clear that people want and need such help from Katrina to Sandy to the mortgage crisis. The issue is that the Buckleyites took their intellectually bankrupt ideas and used a vapid, B-movie actor dumbass to implement them. There was not a modicum of thought about what this might mean for the GOP going forward because they thought that they could rely on the vote of the regressive South and Midwest forever by telling horror stores about liberal boogeymen to those folks.
This is why I think the GOP is dead. They can't extricate themselves from this coalition they have made without being effectively powerless at the national level.
The Democratic party has not done much better at defending free trade, which explains Bernie Sanders, who has half the answer (social programs), but who now has entered the party from the outside and pushing the idea that free trade is a net negativr. Hillary needs to start making the case at some point that free trade is not the issue in and of itself, but that the failure of the government to care for its own in employment and income crises as it would in environmental crises is the problem. When she has more room to continue a pivot to the GE, she needs to gear up to win this argument because Trump is going to have it with her.
It turns out that their voter base finally asked "for whom?" in response to that argument.
I also think that their cleaving to Reagan's philosophy that growth should be distributed mostly to the top, and that social programs to address this are bad and bound to fail, in combination with their support of free trade deals, was always bound to end up with populists trying to take the controls of the party away from the Buckleyites, whose economic philosophy is intellectually bankrupt from the get-go because it benefits too few people.
People laughed at Reagan's "most scary words are 'I'm from the government, and I'm here to help," joke, but it has always been clear that people want and need such help from Katrina to Sandy to the mortgage crisis. The issue is that the Buckleyites took their intellectually bankrupt ideas and used a vapid, B-movie actor dumbass to implement them. There was not a modicum of thought about what this might mean for the GOP going forward because they thought that they could rely on the vote of the regressive South and Midwest forever by telling horror stores about liberal boogeymen to those folks.
This is why I think the GOP is dead. They can't extricate themselves from this coalition they have made without being effectively powerless at the national level.
The Democratic party has not done much better at defending free trade, which explains Bernie Sanders, who has half the answer (social programs), but who now has entered the party from the outside and pushing the idea that free trade is a net negativr. Hillary needs to start making the case at some point that free trade is not the issue in and of itself, but that the failure of the government to care for its own in employment and income crises as it would in environmental crises is the problem. When she has more room to continue a pivot to the GE, she needs to gear up to win this argument because Trump is going to have it with her.
Not really no. Hell most places still ignore Nate and Sam Wang and they've been doing this for years with a good track record.
Got a text from a Bernie supporter yesterday to knock on doors. Either someone gave them my information, or ActBlue shares their data with his campaign. I tossed a donation to Russ Feingold on the day he announced he was challenging Ron Johnson.
1. I don't do primary work for candidates.
2. I don't do GE work for candidates, only issues.
3. I'm voting Hillary if Bernie is still in on April 5.
1. I don't do primary work for candidates.
2. I don't do GE work for candidates, only issues.
3. I'm voting Hillary if Bernie is still in on April 5.
We have no reasons to believe his accuracy will approach 99%.
Wisconsin is going to be fun
TumblingHippo
Member
Bernie, and if not, Trump. A lot of coworkers have that same mentality it seems. They joke about all the racist/fascist stuff he says but if it's him or Hillary they'll vote Trump.
brainchild
Banned
Yeah, but polls also don't have less than 1% margin of error either.
Of course, I'm not saying that Tyler's model would ever get that accurate, but I do think that the level of accuracy matters.
At any rate, if continues to lower his margins in his correct predictions, I think he'll become as relevant as Nate over time, despite not using a single poll, which would be incredible. It would usher in a whole new era of statistical modeling that's never been done before.
Ogrekiller
Member
Not only that media outlets are interested in ratings, not accuracy.
Also what use is a model that becomes more accurate the more the results become irrelevant? His accuracy could get better for all I know but who cares if the nomination is all but over?
brainchild
Banned
Of course not, but this was a hypothetical supposition and B-Dubs said that it STILL wouldn't matter (even if it's not likely to happen in real life), and I strongly disagree with that notion.
This is truly mind boggling to me. I just don't understand. I doubt it's a huge portion of Bernie supporters that will do this but the fact that any of them will is insane. The two politicians couldn't be any more different.
Nate actually loves Google searches for whatever reason and talks about them a lot. Probably his favorite indicator after favorability and endorsements.
Then he runs into the very problem I pointed out earlier: would this even work in another scenario? Unless he can apply this to the GOP or other scenarios with more than 2 people running there's no point. If all he can predict if Bernie vs Hillary then who cares? We already know how that's turning out.
RastaMentality
Banned
Yeah, a lot of my white male friends have been rationalizing possible support for trump and it's kinda terrifying.
The problem with this is that there is still a significant part of the voting population that cannot be adequately counted in this fashion. It's the same issue I expect to start to rear its head when the baby boomers start dying off, assuming telephone polling is still the main polling method, at that point. Assuming his method is in any way viable, I imagine it would be more accurate if he managed to find a way to incorporate traditional polling into his averages.
VendettaRed07
Member
Is this some sort of experiment? Like is he seeing if he can post consecutive career killing gaffes faster than the speed of light or something
- Status
Similar threads
- Locked
- Locked
- Locked
- Locked