Harris vs. Trump : A Data Scientist's Prediction (3 Months To Go)

A data scientist's prediction of the 2024 presidential election with three months to go!

0:00 The Method
13:57 The Results

[mndhamod]

If you place such little faith in the polling data, shouldn't you also put even less faith in the voting data because it is even more outdated? I feel assuming that increasing one means decreasing the other is somewhat flawed. I understand it from a mathematical view, total probability being 100%. But not having faith in polling data should be considered relative to faith in voting data. Thoughts?

[MrMoore]

I will definitely rewatch to help my own understanding, but I would love if you could cover the bayesian updating process in a stand alone video. That concept has always somewhat eluded me :/
Phenomenal work as always!

[gingerderidder]

That's a nice final plot. Will steal it if you don't mind:)

[djangoworldwide]

You're awesome. A question regarding coding style:

Wouldn't it make more sense to either store or just write your uncertainty coefficients outside, maybe in a dictionary or as variables? I feel like keeping them in the actual uncertainty functions (1/4*baseline_uncertinity**2 etc.) is tricky, and allows less flexibility in the longs term.

Would love to hear other people's thoughts

[shshshbbb]

Can you do a similar kind of analysis (with the polling datas) with the previous voting to see the trust level of the polling for that election and apply that to our current analysis?

[ShounakKundu]

Quite a lucid explanation. Thanks a lot for doing this.

[SimonBaier]

Great analysis. I learned a great deal about the analysis process, tools and methodology that results into a reasoned end analysis. I'm an aspiring data scientist, and a keen political observer, and have a couple of thoughts.

First is that results vary widely between polling organizations. Why not add a calibrating function that analyzes historic polling results to eventual outcomes for each of the polliing organizations - the trending prediction as well as the evolving margin of error reported throughout a campaign.

Second thought is the swing states effect - there's a minuscule chance that of the electoral outcome of 40 or so states, so why not focus on the swing states PA, MI, WI, PA, AZ, NV, and for safety NC, GA, VA, NH. It could be reasoned that if Trump loses GA or NC, he loses the election, and the opposite if Harris loses VA or NH.

Thirdly, voting outcomes in these swing states may be better predicted using an alternative methodology to polling. Specifically, the predictive factors of economist Allan Lichtman's 13 keys. he's predicted every election since Reagan. Adding probabilities of the most critical remaining keys would make use of new data, and at very least add tools to weight polling, and increase/decrease margin of error or your existing calculations.

Just some initial thoughts. Keep up the good work.

[EvsEntps]

But does the model account for the significance of the passage of time?

[joshmc]

I wonder what that polling data looks like since Trump flipped Kennedy and Gabbard?

[ccuuttww]

Something u miss is the most vote count
it doesn't mean u win the election! that's the tricky part

[christusrex]

The outcome aligns with my own analysis on the topic, funny enough it seems the election will come down to Pennsylvania which is 19 votes. So it is very likely your model is correct in this case.

[rubyemes]

what would the method have predicted in 2016? I bet it wasn't Trump!