Physics Forums

Physics Forums (http://www.physicsforums.com/index.php)
-   Set Theory, Logic, Probability, Statistics (http://www.physicsforums.com/forumdisplay.php?f=78)
-   -   Predicting Wind Speed based off previous data (http://www.physicsforums.com/showthread.php?t=125129)

Obanion Jul2-06 11:32 PM

Predicting Wind Speed based off previous data
 
Hello all,

My boss wants me to create a form of short term wind prediction based off a collection of long term wind data. I haven't taken a course in statistics so this is totally new to me. Basically, the data is organized by Year > Month > Day > Hour > Wind Speed with data stretching back three years. There is also another set of data that spans over a time period of twenty years. Unfortunately, this data isn't of much use to do the absence of a given altitude which renders it useless for our use.

I was given instructions to take the moving average or something of the sort and try to predict at most 6 hours into the future, but I've never had experience with this sort of thing. I'm limited to using Excel as of this moment, but in a few weeks they will shift a thorough analysis to specialists due to the extreme complexion. This further analysis will include humidity, radiation, and all the other requirements for a decent model. Consequently, my predictions do not have to be extremely accurate, but they should be able to provide a general idea of what the wind speed should be.

I'm quite overwhelmed since I've never had this sort of task assigned before. I've worked with Excel before, but it was mostly for very simple data analysis. I'm looking for guides, methods or suggestions on how I would go about doing this analysis. I was hoping that the users at this forum could be of assistance.

Thanks kindly,

Obanion

I can provide more information

0rthodontist Jul7-06 02:45 AM

One plausible, simple way, you could average all the data by hour of the day (average all 12:00 data, all 1:00 data, all 2:00 data, etc. separately) to get a "standard" curve that the wind speed follows over the course of a day. Then for the current wind speed, take the average of the last few hours and compare it to the average of those hours on your standard curve. Then shift the standard curve up or down so that those two averages match, and use the shifted curve as your prediction. You could also group the data by month--i.e. average all 1:00 PM wind speeds in January for the three years all together, but separate them from 1:00 PM wind speeds in June.

Obanion Jul8-06 08:24 PM

Thanks. I tried out what you suggested but the data is so "fuzzy" that it's almost impossible. I've been checking out some neural network programs and I'm going to try out a few things with them to see how they'll preform.


All times are GMT -5. The time now is 06:27 AM.

Powered by vBulletin Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
© 2014 Physics Forums