IBM splashing $2bn on Weather Company – reports Big Blue is reported to be further beefing up its Watson division by splashing $2bn (£1.3bn) snapping up the digital arm of the Weather Company, a US-based organisation that runs the Weather Channel and a digital company providing mobile apps. Both the Wall Street Journal and the Financial Times reported that people familiar …
Namely that correlating data and identifying trends isn't the same as delivering pertinent and meaningful insights to any given user in an open-ended userbase.
So, in English: The weather forecast will still be shit, but great for extrapolating scary climate change forecasts to back up the hair shirt brigade.
IBM's "Watson" won't improve the "predictions" of the Weather Channel.
Hell I can do a better and more accurate job by just looking at radar.weather.gov and my outdoor thermometer. Here's a hint - GIGO
Watson will just make the "fudged" temperature data easier to swallow by correlating the faked data and ignoring the real data to push us further down the primrose path to "climate change enlightenment".
Now apply the ashes and say it again children "CO2 is bad for you".
The Weather Channel doesn't do climate forecasts they do weather forecasts - 7 to 10 days out at most. More computational power increases the accuracy of those forecasts, but they still get things wrong, especially further out.
They're certainly a lot more accurate for weather five days out today than they used to be for three days out when I was a kid (and the "extended forecast" back then was only five days instead of the 7 or 10 common now) But going from (just throwing out random numbers I have no idea of the real ones) let's say 30% accurate to 60% accurate still means they're 40% inaccurate, which is what people remember - "the weather man said it would be sunny this weekend, and it has been raining all day!"
Is it just me, or have all these weather forecasting apps using big data going backwards?
Five years ago, there were apps like Meteo’s “Weather Pro” that were about 80% accurate, and you could reasonably rely on the forecast for the next 3-5 days. Over the past year or so, none of them are reliable. The only one that you seem to be able to have any confidence in is the next 120 minutes from AccuWeather. I’ve started using satellite radar as the best way to predict whether it will rain, which seems crazy...
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
