Hacker Read

_notreallyme_ · 2023-10-31 03:54:24

You should indeed try and target a 50% accuracy rate.

pachico | karma 1762 | avg karma 2.73 · | 2020-04-24 05:56:34+00:00

Isn't 70% accuracy too close to random (50) to say it works?

peteforde | karma 9134 | avg karma 6.25 · | 2024-02-20 00:01:12

Is 90% accuracy good enough to be useful?

yvdriess | karma 1325 | avg karma 2.71 · | 2023-09-20 11:13:30

What do you mean with accuracy here? Usually 50% accuracy means cointoss, meaning 20% accuracy is equal to 80% accuracy, which is better than the article's 78% and not that far from 90%.

bobwaycott | karma 3438 | avg karma 1.8 · | 2013-08-15 21:32:13+00:00

I find it difficult to square a 10-40% chance of success with "fairly good accuracy".

jjice | karma 4230 | avg karma 3.39 · | 2022-01-06 14:32:59

If my accuracy was 76%, I'd be upset. I'd definitely work on accuracy over raw speed at that point.

andrei_says_ | karma 6527 | avg karma 2.62 · | 2023-11-28 13:47:41

So, 70% accuracy with 100% confidence?

causi | karma 10384 | avg karma 4.36 · | 2023-04-21 07:11:57

I'm willing to bet your accuracy was better than 4% across an entire field.

joshdance | karma 2513 | avg karma 3.16 · | 2013-06-05 15:52:45+00:00

Love this idea. However 94% accuracy is not good enough. You need to be able to trust your action will lead to the desired result. 94% is like Siri getting your whole text message right, but getting the recipient wrong.

usea | karma 1896 | avg karma 4.76 · | 2012-07-17 01:04:37

Even for 0% accuracy?

thih9 | karma 6590 | avg karma 2.78 · | 2023-02-21 11:30:14

I don't think that was the point. I think the point was that as long as you're happy with 90% accuracy, you can have that for 10% of effort.

It requires a perspective shift. Then again, cost efficiency is a big motivator.

reply

luckylion | karma 3881 | avg karma 0.72 · | 2020-09-01 14:45:22+00:00

Nobody needs 100% accuracy though, they just need better-then-current accuracy.

lurker14 | karma 606 | avg karma 2.05 · | 2012-08-27 15:16:30+00:00

100% accuracy does not exist in real-world systems. 99.99999% accuracy might be good enough, though.

victor9000 | karma 2144 | avg karma 5.48 · | 2023-06-08 08:39:30

Well, how else are they supposed to reach 99% accuracy!?

_delirium | karma 42153 | avg karma 4.32 · | 2011-05-13 13:27:29+00:00

If they're currently getting 79% accuracy, there must be some significant signal there. No idea if 95% is achievable, but unless they're already using the absolute optimal approach, something at least into the 80s should be possible.

fla | karma 2699 | avg karma 5.11 · | 2014-01-09 13:34:05+00:00

I really hope they mean 99% accuracy.

Kluny | karma 4287 | avg karma 2.57 · | 2021-03-06 01:38:33

Sure, you could probably train an AI to get 72% accuracy on that as well. It would be just as meaningless.

nawitus | karma 3654 | avg karma 2.15 · | 2014-07-11 15:06:57+00:00

It's just a thought experiment. If the accuracy is -+n%, then lets say "2*n% worse".

callesgg | karma 1359 | avg karma 1.35 · | 2021-03-05 21:20:11+00:00

Hum i take this back the calculation should be 0.13 * 0.89 + 0.87 * 0.5 = 55% total accuracy

probably_wrong | karma 8074 | avg karma 5.61 · | 2018-10-04 05:48:49

Accurate? Sure. But accuracy is only one part of the picture: a system with 32% accuracy but a 0% rate of false positives would be very useful.