Talk:Probabilistic forecasting

Latest comment: 4 years ago by Vermorel in topic Supply chain use

Only weather?

edit

The lead describes the topic as "...a technique for weather forecasting...". Much of the rest of the article also talks about weather forecasting. But is its use restricted only to weather? If it is weather only then we should remove non-weather items. But if it is general (and weather is merely one, amongst many, of its applications) then we should edit the article to make clear this generality. Feline Hymnic (talk) 19:35, 2 January 2011 (UTC)Reply

Indeed the article does appear to me to only apply to weather, but that just seems flat-out wrong to me. Certainly predictive models from all sorts of domains are free to give outputs in terms of probabilities rather than discrete events. I don't know the terminology well enough though to make definitive edits. Mickeyg13 (talk) 19:15, 7 June 2011 (UTC)Reply

It seems "terminology" is not the right question, rather it should be finding actual examples of (important) use in other types of application. Melcombe (talk) 08:37, 8 June 2011 (UTC)Reply
Well both gambling and the stock market both involve trying to predict the probability of some event(s) occurring. When Vegas lists certain types of betting lines there is an implicit probability defined by the odds. I don't know the terminology to know if these examples count. The article makes it seem like the only predictive models in the world whose output is a probability must be talking about weather. Mickeyg13 (talk) 17:27, 14 June 2011 (UTC)Reply
I am not sure that I would count the gambling context here, but maybe. I think the article is about cases where the probabilities assigned change given known information, so that proabilities of cards or hands that are fixed wouldn't be covered, but probabilities of ending up with a certain hand given cards already held or seen might just count ... but I think such specialised stuff would be better elsewhere. For non-weather context, the article already has "probabilistic forecast of energy usage" in the Examples section. Perhaps once more examples are found the start of the article cdan be rewritten. I think one of the finance-oriented articles has something relevant to do with forecasting economic inflation. Melcombe (talk) 09:07, 15 June 2011 (UTC)Reply
Melcombe, I wasn't speaking strictly of table games but more of sports betting. If Vegas lists the moneyline odds of Team X beating Team Y as say -200, there is an implied probability that Team X should win no more than about 66.67% of the time. If a professional gambler uses a different forecast that says Team X should win 75% of the time, he would expect to win money by placing a bet. To me that definitely seems like probabilistic forecasting, but as I said I don't know the terminology well enough.
One of my gripes with many statistics (or related) articles on Wikipedia is that oftentimes the same concept applies to several domains, but different domains have different names for it (e.g. positive predictive value looks like precision to me). It's often not clear whether the term in question is domain-specific or not. I don't know the domains well-enough to change the articles, and multiple requests for clarification on talk pages have been ignored. Mickeyg13 (talk) 19:53, 30 June 2011 (UTC)Reply
I have loooked at the associated weather-related articles and I think that there is enough in those about probabilistic forecasting that this article need not contain much more than it does, leaving scope for exapansion to other applications. I have rewritten the lead to start more generally than it did, leaving scope for others to add new sections on other types of application (including gambling). Hopefully these can be brief and provide pointers to existing Wikipedia articles. And the lead should be changed as appropriate. Melcombe (talk) 12:42, 1 July 2011 (UTC)Reply

Supply chain use

edit

Disclaimer: (1) I am the author of the page listed below. (2) My company is selling a supply chain software.

Probabilistic forecasting are of prime interest for supply chain purposes. See https://www.lokad.com/probabilistic-forecasting-definition I believe that this page is worthy of being referenced in this article, but naturally due to the obvious conflict of interest, I will let someone else proceed if deemed relevant. Joannes Vermorel (talk) 09:30, 11 November 2020 (UTC)Reply

How meteorologists define probability

edit

If I Google this term, this Wikipedia page is the top hit. When I go to the hour-by-hour page of the Weather Channel website, I might learn that the probability of rain is 30% at 2:00, 40% at 3:00, 60% at 4:00, and 50% at 5:00. On the daily page, probability of rain for the entire day is shown as 60%. This is not how probabilities are calculated. It would be helpful if this page included a sentence explaining how these probabilities are assigned. — Preceding unsigned comment added by Rmrichman (talkcontribs) 14:15, 4 April 2014 (UTC)Reply