Archive for April 5th, 2013


How accurate are your forecasts???

I remember when I was in college, participating in the Forecasting Contest…against other students, grad students, and professors. And I finished woefully as I recall, mostly because they knew that to score well (whether you decide this strategy decreases the value to you is another story), they had to play odds, hedging on percentages, and almost never going 100% or 0% chances. I was a more black and white forecaster, because IN MY MIND, it either rained or it did not…it never 70% rained! Right? However, over time, I learned there is some value in their methodology, because while I’d want to be as precise as possible for YOU the viewer, weather is NOT an exact science, and by telling you words or phrases like “likely, not likely, slight, most, probable” etc, these DO convey to the viewer my level of confidence in my own forecast. This has value because if I am definite, you can therefore BOOK it. If I have serious doubt, you should play your weather-related activities by ear or even wait until I can be more certain.

Some years later, I did my own (personal) study…for my own curiosity, I wanted to see what I’d score on a 5 day forecast. It required months of forecasts and then lengthy reviews and analysis, and I came up with an 82% overall accuracy rate, for myself, at that time (maybe 12-15 years ago, now. I’d like to think I’ve improved even more since…just due to time served!)

In any event, I remember when I graduated college in the late 80s, serious studies had shown that the most accurate forecasts/forecasters would be in the 80-90% range day 1 and 2, in the 70s day 3, and the accuracy trailed off to about 50% by day 5. Now think about that…if the accuracy of day 5 is 50%, that is 50/50, and that means…at that time…that if someone told you a forecast for day 5 only, and you flipped a coin yourself for Day 5, the mindless coin would be as accurate (over time) for Day 5 as the person who is studying all the weather charts…boring through material…trying to give you an accurate day 5 forecast! That’s not encouraging for us Mets (Meteorologists)!

Well, all the above is a prelude to a new study, demonstrating that the accuracy overall has improved to the point where, not only is Day 5 better than 50/50 now, a current day forecast for days 6 and 7 are now as accurate as an old “5th day” forecast…as the first 5 days of accuracy have improved to the point of demonstrating “skill” even through day 5 now, so that days 6 and 7 are now worthy of benefit to the receiver.

Check it out:

PS: As a footnote, I would point out that forecasts…or any information relayed…is only as accurate as the viewer’s memory. In other words, I cannot tell you how many times I have given a specific forecast, only to have a viewer tell me the next day in the store that I was wrong when I said this or that…AND I NEVER SAID what they repeated back to me! So, the ATTENTION-span of the viewer, the RETENTION-span of the viewer are also equally important to the actual forecast itself. It’s a fact that if I tell a person the high temps for the next 5 days, they will almost never repeat them back to me correctly, or in the right order. So that’s a huge part of the misperception by some, that “You Mets are never right!”