about forcasting with moving average
 

Do you know what might be a problem with forcasting with moving average?

No, we don't...especially since you provide no details/context AT ALL to let us know what you're even talking about. Read the "Question Guidelines" link in my posting signature.

well i'm sorry, actually i'm doing my research about forcasting with moving average, we are developing an inventory system with forcasting with moving average.. i just want to know what might be the possible problem with it. thanks

Again, you are providing ZERO details/context for your question; restating it still tells us absolutely nothing. This is a Linux related forum, so start with telling us how, exactly, your question is Linux related? Again, read the "Question Guidelines" link in my posting signature.

You need to tell us what version/distro of Linux you're using, what piece of software, what the actual error(s)/problem(s)/message(s) are that you're getting, etc. If you have none of that, there is nothing we can tell you or help you with. If you want to do research on a particular model, there are many other sites that can help.

As you've been told, "we're developing a forecasting system with moving average ... what might be the trouble with it?" is an unanswerable question which would also get you strange looks in a college statistics class.

From WikiPedia: There are no "problems" associated with the use of such an algorithm, only "implications." If you are uncertain about the statistical bases of your proposed forecasting system, I suggest that you visit your local University and discuss with a qualified statistician. A forum such as LQ cannot help you in any meaningful way.

As Mark Twain quipped, "there are lies, damned lies, and statistics." A statistical algorithm is worse-than useless to you if you don't understand what it means and if it doesn't match your real-world situation and data. Someone on the project needs to know what statistical methods are to be used, and why. In addition, you should seek to use "known good" implementations of statistical algorithms.

Hi vhen_marin and welcome to LQ.

There are a variety of formulas which you can apply for a moving average. Have you picked one or more which you are considering? And why do you feel there can be possible problems related to Linux?

Are you programming an algorithm on a Linux system using a particular language?

As part of my job I write signal processing filter code from time to time, some of that involves things like a boxcar filter or maybe a CMA algorithm, and other filters such as FIR or IIR.

To me, they are all just a computer implementation of a mathematical filtering, or signal processing, method. I'd also like to note that many moving average algorithms are well documented.

Therefore the requests on the part of LQ members here should be very reasonable, and you should perhaps elaborate a bit more on what you're doing. Please tell us if there is a normal model which you feel you intend to forecast with and how you intend to accomplish the actual implementation for this.

Please also realize that I'm trying to evaluate if this is a general discussion, a Programming discussion, or a Linux question.

In general, LQ does not recommend you use an actual email address as your username, it can lead to potential spam or advertisements, not from LQ, but from web search results showing all LQ questions where your username shows up.

To request a change in your username, please use the Contact Us link here, or seen also at the bottom of every LQ page form.

Best Regards.

The problem with simple moving averages can be if the data contains large blips - these then stick in the average until the end of the window when they drop out changing the moving average at that point even though there was no significant change in the incoming data at that point. Consider an exponential decay smoothing ( https://en.wikipedia.org/wiki/Exponential_smoothing ). It is easy to implement and gives more intuitive results that a simple moving average.