PDA

View Full Version : Small mistakes?



djones520
12-04-2010, 11:57 AM
Asa weather forecaster, I often times get things wrong. When having to take into account the actions of the entire atmosphere, some small things are going to fall through the crack.

Temperatures for one. It's cold here. We have a massive front roll through yesterday, and it brought down an extremely cold air mass with it. Dropped our temps to -14c yesterday morning.

Our models seemed to indicate that the majority of the cold air advection would have been in that first 24 hour period, and we'd start to see a gradual rise in temperatures by a couple degrees a day, with yesterday being the coldest day. This means we were thinking we'd see a low temp today of maybe -12c.

Well... I just checked the temperature, and it's -21c outside right now, with clear skies and about 8 hours to go before sunrise, it's a safe bet we could see -25c.

Oops.


So, what are some other little errors that you guys make in your work that don't really have resounding impacts but you still feel dumb about doing?

Rockntractor
12-04-2010, 12:04 PM
In heavy equipment work. there are always little errors, we generally have to redo or repair all of them.
Luckily our mistakes don't get televised unless there is a sinkhole in a road or something like that.

Madisonian
12-04-2010, 10:03 PM
2 material handling gantries, common beam, common pick and place, oops programming, lotta noise.:o

patriot45
12-04-2010, 10:06 PM
Sometimes I put too much salt in my recipes but I can fix it by throwing in a peeled potato!

I hope that helped.

Gingersnap
12-04-2010, 10:19 PM
In modeling, we usually get blindsided by assuming that past performance is a strong indicator of future performance. Usually it is but frequently we miss trends that impact our models even though they seemingly have nothing at all to do with our expected inputs.

We neglected to factor in the fact that many illegal aliens drive old, non-compliant junker cars. A few hundred of these cars have no impacts, many thousands sadly did and it screwed up our estimates and projections. We have nothing at all to do with immigration or social issues but this is one we kicked ourselves for overlooking. :(

megimoo
12-04-2010, 10:51 PM
In modeling, we usually get blindsided by assuming that past performance is a strong indicator of future performance. Usually it is but frequently we miss trends that impact our models even though they seemingly have nothing at all to do with our expected inputs.

We neglected to factor in the fact that many illegal aliens drive old, non-compliant junker cars. A few hundred of these cars have no impacts, many thousands sadly did and it screwed up our estimates and projections. We have nothing at all to do with immigration or social issues but this is one we kicked ourselves for overlooking. :(Modeling is always tricky business .The interaction between models during an simulation and the resulting predicted response always need to be refined .

The more dynamic the stimulus the greater the departure from the predicted.Something like accurate atmospheric modeling is a pure pipe dream.There are far too many unknowns to build an accurate model and the unknown atmospheric interactions are enormous.

The most accurate modeling is done at Dream Works where they have run time model in a custom Cmos Cray Processor with its response to a given stimulus already on line and ready for the main processor to accept their data .

Each Character has it's own model and it's own Cray.The simulations some times run for an entire week if the models to be stimulated are plentiful and the interactions are in high resolution.

With a simulation changing with time the problem becomes much more profound.The model time stamps are now dynamic in nature and the model itself has to change to accommodate the dynamic stimulus changing in time .

NJCardFan
12-04-2010, 11:56 PM
September 14, 1991...wait, that was a big mistake.

Kay
12-05-2010, 12:15 AM
In my line of work (real estate) there is always that fear that you're going to get right down to the closing table and have something pop up like an error in your estimated closing costs. You thought the survey was going to run about $2,000 but the surveyor doesn't tell you he ran into extra issues and you see the bill right before closing day is actually $4,500 and have to explain that extra $2,500 to the buyer. :o

djones520
12-05-2010, 10:16 AM
In modeling, we usually get blindsided by assuming that past performance is a strong indicator of future performance. Usually it is but frequently we miss trends that impact our models even though they seemingly have nothing at all to do with our expected inputs.

We neglected to factor in the fact that many illegal aliens drive old, non-compliant junker cars. A few hundred of these cars have no impacts, many thousands sadly did and it screwed up our estimates and projections. We have nothing at all to do with immigration or social issues but this is one we kicked ourselves for overlooking. :(

Yeah, we have the same concept, call it persistance. I know a lot of lazy forecasters who just use it. They'll just look at the day before and say thats what is going to happen today. It'll usually work, but on those days where it doesn't things get way off.

Rockntractor
12-05-2010, 10:18 AM
Yeah, we have the same concept, call it persistance. I know a lot of lazy forecasters who just use it. They'll just look at the day before and say thats what is going to happen today. It'll usually work, but on those days where it doesn't things get way off.

Those are the days I give you a hard time about.:D

SaintLouieWoman
12-05-2010, 11:57 AM
My biggest potential mistake was actually a happy one. I sold a line of equipment that was more expensive than my competitors'. In 1995 I almost didn't complete a very long, demanding bid with the City of St Louis, thinking that it was pointless. My boss procrastinated giving me the final authorization of pricing, so I stayed at the office til almost 9pm and delivered the bid the next day with only 5 minutes to spare (had to be time-stamped on delivery. If not on time, the bid wouldn't be accepted).

I almost didn't stay for the formal bid opening, figuring I didn't want to be embarassed with high losing numbers. I decided to chance it, so put more money into the parking meter and returned to the bid room.

Slowly I realized that I would be one of the 4 winners. I couldn't believe it. I somehow held on to that business for 15 years. It became my largest account. Many of the customers to this day are still my friends.

So sometimes you cannot always predict the future based on history. That's when I learned to never give up----on anything.

megimoo
12-05-2010, 12:17 PM
Those are the days I give you a hard time about.:D

http://www.youtube.com/watch?v=QhzbzwPNgXA&feature=related

The Days Forcast by Mr Jones :
http://www.youtube.com/watch?v=MhPu5AHDMHM&feature=related

RobJohnson
12-05-2010, 12:42 PM
In modeling, we usually get blindsided by assuming that past performance is a strong indicator of future performance. Usually it is but frequently we miss trends that impact our models even though they seemingly have nothing at all to do with our expected inputs.

We neglected to factor in the fact that many illegal aliens drive old, non-compliant junker cars. A few hundred of these cars have no impacts, many thousands sadly did and it screwed up our estimates and projections. We have nothing at all to do with immigration or social issues but this is one we kicked ourselves for overlooking. :(

I knew you were a runway model and all that chemistry stuff was just a front! I'll be having good dreams tonight! :D