Data science and predictability of outcomes.

[minestrone]

Why have our resident data scientists not predicted brexit?

[BrilloPad]

They did not ask Mystic Meg?

[DimPrawn]

Why have our resident data scientists not predicted brexit?

Because it's hard to predict crazy black-swan events?

[minestrone]

So, I work with people who spend more time telling you about their ability and qualifications on machine learning than making accurate predictions.

I'm kind of going to say much of this is pseudo science.

[greenlake]

[scooterscot]

There's deterministic and probabilistic. Deterministic requires *real* facts with which to work with. Since they were not presented during the campaign...

So we're left with probabilistic. Probabilistic requires some similar past event with which to make a prediction. This was the first EU referendum we've had so... we are where we are.

[vetran]

So, I work with people who spend more time telling you about their ability and qualifications on machine learning than making accurate predictions.

I'm kind of going to say much of this is pseudo science.

you are obviously a cretin





love ASS

[jamesbrown]

There's deterministic and probabilistic. Deterministic requires *real* facts with which to work with. Since they were not presented during the campaign...

So we're left with probabilistic. Probabilistic requires some similar past event with which to make a prediction. This was the first EU referendum we've had so... we are where we are.

No it doesn't. Deterministic vs. probabilistic is a modelling choice (physical processes aren't fundamentally uncertain at macroscopic scales). Probabilistic predictions don't require historical data any more than deterministic ones. A conditional prediction may be better than an unconditional one.

[scooterscot]

Where did I say probabilistic prediction requires historical data? I said it requires a similar event. You could create such an event with which to make your prediction. Stochastic processes don't care for time.

Deterministic prediction never makes use historical data either. That's a major slip up in education that is. They measure the likelihood of events through measurement of some quantity.

[teapot418]

I read tealeaves. Shout if you want yours doing.