European Experts Exchange (EEE.org) :: Free Experts PHP MySql Oracle Microsoft Web SQL Delphi


Algorithms :: Maths :: statistical data
By: lneilson	Date: 12/07/2003 00:00:00	Points: 125	Status: Answered Quality : Excellent
hi everyone I have years of daily, monthly, and yearly statistics is there a formula or algorithm that will allow me to predict next monthes figures . thanks lneilson
By: VGR	Date: 13/07/2003 06:25:00		Type : Answer
a lot : -time series -linear extrapolation -polynomial extrapolation -lesser squares extrapolation -(I don't know the translation) "moyennes mobiles" (moving averages ? ) -etc one of the simpliest is to compute a trend analysis on last month's daily data to predict the (few) next days' values, on previous year's monthly data to predict next month's value, etc you got the picture.

By: lneilson	Date: 13/07/2003 09:18:00		Type : Comment
got the picture?? not at all remember some of us out here are mathdiots please explain further lneilson

By: GwynforWeb	Date: 13/07/2003 12:22:00		Type : Assist
This is a big question without a simple answer and it depends on the data and what you are trying to do with it. Ideally we would need to see some typical data. This topic has been discussed before have a look at <A HREF="http://www.experts-exchange.com/Miscellaneous/Math_Science/Q_20546632.html">http://www.experts-exchange.com/Miscellaneous/Math_Science/Q_20546632.html</a> Cheers GfW

By: higginspi	Date: 14/07/2003 01:09:00		Type : Assist
Hire a statistician or buy some statistical textbooks and read them. WAY too much info to teach you about it here.

By: wytcom	Date: 14/07/2003 13:55:00		Type : Assist
You may be able to use a neural network. A neural network can be used to predict future results based on expected conditions when you have lots of data to show past results for similar conditions. Search the web and you will find plenty of information on neural networks including downloadable software to work with.

By: grg99	Date: 15/07/2003 21:43:00		Type : Assist
Well, basically impossible, as we have to access to the future. But if you make some ASSUMPTIONS, many of them rash, like "Sales are going to continue on this trend", or "Sales are going to follow the average slope fo last-year's curve", then you can make some mathematical extrapolations. For example, if you asume sales are going to follow last month's trend, (somewhat unlikely, but often used), you just add the change in last months sales to this months, to predict each succeeding month. This is often doen by research firms that charge many thousands of dollars for their reports, so it's a honorable or at least passable method. or if you ASSUME sales are going to follow some yearly seasonal variation, which isnt a bad assumption if you're in some kind of seasonal business, like tourism or farming, you could take last years figures, and scale them proportionally to the numbers you have for this year. Keep in mind any assumptions you make are dubious at best, but that's life.

By: higginspi	Date: 15/07/2003 22:31:00		Type : Comment
You could try regression models if you think think your sales depend on certain data. For example, you could use month or quarter as a dicrete variable to predict sales.

By: GwynforWeb	Date: 15/07/2003 23:05:00		Type : Comment
lneilson The method one would use is very dependent on the data and how far into the future you want to go. As I said before post some data, (add a constant to it if it is properiety to disguise it), and the group will probably be abled to find the right method for you. The answer depending on your data and requirements could be very simple. GfW

By: TheFalklands	Date: 15/07/2003 23:22:00		Type : Assist
Do research on Decision support systems.

By: lneilson	Date: 15/07/2003 23:25:00		Type : Comment
hi experts thanks for all the replies and I am narrowing down the solution GfW I tried to post my excel spread sheet but it looked scrambled how can I post it? lneilson

By: GwynforWeb	Date: 15/07/2003 23:43:00		Type : Comment
lneilson Try selecting the columns and then copy and paste into notepad first, then copy and paste in EE comment box. GfW

By: oktexasguy	Date: 16/07/2003 00:40:00		Type : Comment
fyi- you can also "save as" and choose text as the type

By: KenAdney	Date: 16/07/2003 02:52:00		Type : Comment
I think Excel uses the least squares method for extrapolation. For the woefully ignorant, you can use Excel to extrapolate trends. Highlight the present figures and drag the box by the little black square in the lower right corner. It'll fill in the next squares in the same way that you can start with January, February, March & have it continue to fill in the rest of the months of the year. Although this works, I have never found it to be accurate.

By: GwynforWeb	Date: 16/07/2003 04:08:00		Type : Comment
lneilson The straight drag drags to a linear regression, there are a number of other forcasting tools on Excel and one of them may be the one you want in which case you problem is simple. Look up 'trends' in the help index and play with a few of them to see if one of them gives you results that intuitively seem right. My sense is you may want a moving average type extrapolation. GfW

By: VGR	Date: 16/07/2003 06:38:00		Type : Comment
same sense for me :D

By: leojl	Date: 16/07/2003 16:28:00		Type : Comment
yes leo

By: higginspi	Date: 17/07/2003 00:27:00		Type : Comment
VGR, >>-(I don't know the translation) "moyennes mobiles" (moving averages ? ) In English it is called moving averages! :-)

By: VGR	Date: 17/07/2003 04:01:00		Type : Comment
I knew it :D I just wanted to play the game the humble way :D

By: Mutley2003	Date: 26/07/2003 16:50:00		Type : Comment
Is this question still alive? If so, can you please clarify - whether you indeed want "a simple formula" - or an existing app - or source (?in what language) If you want a forecasting app, try freefore at <A HREF="http://www.autobox.com/freefore.htm">http://www.autobox.com/freefore.htm</a> - this will at least give you a feeling for the problem and the nature of the data at hand. If you want source, you can find plenty of free source code on the web but you will at least need to know what algorithm you wish to apply. If you are a beginner, I would suggest you do not use ARIMA type approaches, but stick to something like simple exponential smoothing, or exponential smoothing with trend. These forecasting approaches have been shown in forecasting competitions (eg the M3 competition) to achieve 'good accuracy' on average, compared to other more complex approaches. But the best approach does depend very much on the data and the purposes to which the forecasts are to be put. If the forecasts are in any way 'mission critical' then you need to be VERY careful about how you go about the process. I urge you to visit the "forecasting principles" website <A HREF="http://www-marketing.wharton.upenn.edu/forecast/welcome.html">http://www-marketing.wharton.upenn.edu/forecast/welcome.html</a> and review the checklist of approaches .. read the reviews of the "Principles of Forecasting" handbook. The site also has links to forecasting software. There ARE good books on forecasting .. "Practical Forecasting for Managers" by Nash and Nash is very accessible and uses Minitab throughout. Also "Forecasting - Methods and Applications" by Makridakis, Wheelwright and Hyndman is very readable. At a more advanced level, perhaps go to "Time-Series Forecasting" by Chatfield. Lastly, if you really just want formulae there is quite a good summary of "equations for the smoothing models" (includes forecast and variance of prediction errors) at <A HREF="http://www.csc.fi/cschelp/sovellukset/stat/sas/sasdoc/sashtml/ets/chap30/sect11.htm">http://www.csc.fi/cschelp/sovellukset/stat/sas/sasdoc/sashtml/ets/chap30/sect11.htm</a> These are the formulae used by SAS, I believe.

By: OpConsole	Date: 28/04/2004 02:14:00		Type : Comment
This question has been classified as abandoned. I will make a recommendation to the moderators on its resolution in 4 days. I would appreciate any comments by the experts that would help me in making a recommendation. It is assumed that any participant not responding to this request is no longer interested in its final disposition.

By: KenAdney	Date: 28/04/2004 02:21:00		Type : Comment
I think the question has been answered & this is worth retaining

Do register to be able to answer

©2010 These pages are served without commercial sponsorship. (No popup ads, etc...). Bandwidth abuse increases hosting cost forcing sponsorship or shutdown. This server aggressively defends against automated copying for any reason including offline viewing, duplication, etc... Please respect this requirement and DO NOT RIP THIS SITE.
Please DO link to this page! Post to Plurk Post to Yahoo Buzz Post to Delicious Post to Digg Post to MySpace Post to Ping.fm Post to Reddit Post to StumbleUpon

Contact

page generated in 342.985870 milliseconds

Why Google AdSense ads ?

New Improved Search!