Open office trend line
Add an attachment proposed patch, testcase, etc. Ooo speranza m trading You need to log in before you can comment on or make changes to this issue. Description Unknown UTC I sat down to do my Physics homework, and was disapointed to find that I had to restart the dreaded excel in order to use a second degree polynomial trendline.
I would like you to add the feature enabling the use of x degree polynomial trendlines. Please note that Issue gives clear details about this feature. Please find enclosed another version which try to correct the problem. I did also my best to translate it in english. Comment 10 jumbo UTC Created attachment [details] New version of CorelPoly which correct correlation coefficient Comment 11 hugovazquez UTC This is absolutely a basic feature, i don't understand why there are more advanced regression types such as logaritmepic or poware when a polynomial regression is lacking.
Besides, the rest of the regression are very open office trend line. There are no options to manually adjust regressions.
This features must be added in the next realeases, it is basic. Comment 12 jumbo UTC Hello, hugovazquez : For me it's must have feature, as well for engineering. But saying that would never help programmers.
At least you should vote for this issue. I think this issue should have more than 9 votes. To help going a little forward: this feature requires to modify dialog box of Regression Curve and Statistics, as for "moving average" regression type issue These two issues may be resolved simultaneously.
The minimum to do to be compatible with MS-Excelis to have the order of the polynom default should be 2 as open office trend line in issue To go farther, minimum default 0 and maximum orders should be given, as in the macro of Laurent Godard see attachment. Exponential terms are useless for me, but may be Laurent Godard has examples where it is necessary.
All these values require to be added in the dialog boxes. One problem which may occur during calculation of coefficients, is a non inversible matrix. I don't know which conditions are necessary to avoid this problem. Another problem in case of negative orders, will be with null values of x.
Hope it could open office trend line resolving this feature I noticed that all other types only need two variables to be calculated. Is it the stopper point of this issue? For people who use spreadsheets for 'everyday' quantitative correlations, the lack of polynomial regression equations in OOo Chart is an obvious deficit.
I'm now convinced that Sun should send their programmers off for a week to work with users so that they can see for open office trend line how spreadsheets are used.
Hopefully, this will motivate them to work on things that are fundamentally missing. This is not the only issue that is most important to users.
Documentation/How Tos/Calc: TREND function - Apache OpenOffice Wiki
It is just a matter of fact that I cannot solve them all at the same time. If you have developing skills, you might want to help?
Also a specification or a file format change request would be helpful. And yes, please use your votes! It makes arguing easier. This can only mean they are working on things that the "users" don't value as much. In product development, it is imperative that the design team do market studies and test group studies to find out what is important to the customers.
Scientists, engineers, managers, and most others want to learn from that data which means applying statistics and trendlines to that data. This is a core feature, no exceptions.
I hope to graduate in about weeks. Let me know if you are serious about your willingness to work with me on this.
- How to make money on the Internet in dollars
- Changing the chart type You can change the chart type at any time.
I can help with the curve fitting theories but probably not the actual implementation of the code. I've offered my help before on trendlines and it wasn't taken. If the theory is missing or unknown, then that is where I can help. Again, I'm tied up for about weeks but can point a programmer in the right direction for learning about curve fitting to minimize error An algorithm would be a first step.
Input would be a sequence of x and y double values and the requested order open office trend line the user wants to decide that, maybe in addition a minimum exponent? Output would be a sequence open office trend line coefficients and exponents somehow. Some of the input values might be NAN or not finite, that must be taken into account option what does it mean. The algorithm must be independent from spreadsheet access as it must work also for charts from writer tables or charts with own data for example in impress.
And good luck for your examination. Comment 24 jumbo UTC Hello, In order to not invent again the wheel, there is a macro in the attachment of this issue. It is written in OOo Basic, and algorithm is quite straightforward. Comment 25 jdpipe UTC Please could this be brought forward? The maths involved in implementing this open office trend line is really not that hard, and it's a very useful feature provided by What business to earn quickly and a lot that many scientific users will immediately notice.
Comment 26 jdpipe UTC Please could this be brought forward? Also, I'd add that this feature needs to permit the use the select the order maximum power of 'x' that they want to have in their polynomial regression. I am open office trend line to say, that I still have not the time to look at this issue more deeply. And still there is the problem with the file format! This feature cannot be saved with ODF1.
Best would be to implement this feature, thus it is known what fileformat changes are really needed. And then integrate the feature into the master codeline. If someone out there wants to implement this feature just drop me a note and start!
Help is welcome! Comment 31 aveldro UTC As a mechanics engineer, I often use polynomial trendlines degree 2 open office trend line generally enough on MS XL at work, to modelize xy data. I consider that this feature is necessary! Comment 32 jumbo UTC aveldro : as mentioned above, it is useless to ask for a feature if you do not vote for it.
With only 36 votes, it is less than 41 votes for bubble charts for instance. So developers may focus only on what is more asked Obviously, it didn't work. It is done now.
Linear Regression with OpenOffice
I am surprised that this is not considered as an important feature. Gnumeric has it for instance.
Comment 34 jumbo UTC Hello, Gathering open office trend line about polynomial trend line, I noticed that again solution is in front our eyes.
An easy way to get polynomial trend line equation is given in Help of trend line dialog box : simply use LINEST function, with proper construction of X data. From Help : The polynomial regression equation A polynomial regression curve cannot be added automatically. You must calculate this curve manually. Comment 35 fjcc UTC Thanks to jumbo for pointing out the work-around, but it does not, in my opinion, solve the problem.
The method is far more complicated and slower than clicking a few boxes in a dialog, is prone to error how open office trend line times will I forget that the coefficients are listed in reverse order!? In my workplace, the technicians who work up data would find this a poor substitute for a polynomial option in the Add Trend Line dialog. It will, however, be a real open office trend line for many people to have this method available.
Comment 36 jumbo UTC Hello, I am looking for help about user experience with polynomial regression curve.