[graphics-issues] [Issue 20819] add polynomial regression type
To comment on the following update, log in, then open the issue: http://www.openoffice.org/issues/show_bug.cgi?id=20819 --- Additional comments from hedleyro...@openoffice.org Thu Sep 23 20:59:21 + 2010 --- @jumbo444 a) You need upwards of 9 SF in many of the coefficients to achieve a decent fit. Realistically, we need 11 digits for this case, which is a lot when inverting a poorly-conditioned matrix using brute-force methods. b) The fact that even the algorithm used at XUXU* works reasonably** if you transpose X to be symmetrical about zero (i.e. -5, -3, -1, 1, 3, 5) may be indicative of other possible useful techniques. *I notice that for moderately large zero order coefficients, XUXU truncates zero-order coefficient at the decimal point even when the resolution is much lower than for the other coefficients - is this indicative of other strange behaviour? **But even here adding third-order fits to successive recorded errors provides successively better fits - I assume that this indicates significant residual issues with the algorithms? - Please do not reply to this automatically generated notification from Issue Tracker. Please log onto the website and enter your comments. http://qa.openoffice.org/issue_handling/project_issues.html#notification - To unsubscribe, e-mail: issues-unsubscr...@graphics.openoffice.org For additional commands, e-mail: issues-h...@graphics.openoffice.org - To unsubscribe, e-mail: allbugs-unsubscr...@openoffice.org For additional commands, e-mail: allbugs-h...@openoffice.org
[graphics-issues] [Issue 20819] add polynomial regression type
To comment on the following update, log in, then open the issue: http://www.openoffice.org/issues/show_bug.cgi?id=20819 --- Additional comments from hedleyro...@openoffice.org Tue Jun 22 21:32:16 + 2010 --- Still waiting, but using LINEST meanwhile. The problem with this is that using it for real work clutters up the spreadsheet horribly. However, as I often want to plot the difference between the derived formula and the data, it does have some advantages. The optimum for this would be to be able to add the fitted equation directly into a single column of the spreadsheet. Would this be subject to the same sort of issues as creating a higher-order trendline directly on a graph? - Please do not reply to this automatically generated notification from Issue Tracker. Please log onto the website and enter your comments. http://qa.openoffice.org/issue_handling/project_issues.html#notification - To unsubscribe, e-mail: issues-unsubscr...@graphics.openoffice.org For additional commands, e-mail: issues-h...@graphics.openoffice.org - To unsubscribe, e-mail: allbugs-unsubscr...@openoffice.org For additional commands, e-mail: allbugs-h...@openoffice.org
[graphics-issues] [Issue 20819] add polynomial regression type
To comment on the following update, log in, then open the issue: http://www.openoffice.org/issues/show_bug.cgi?id=20819 --- Additional comments from hedleyro...@openoffice.org Thu Feb 11 10:27:50 + 2010 --- Unfortunately, my use of non-open-source software is anything but a threat. I find myself compelled to use this in order to operate reasonably effectively. I expect that jdpipe is correct that fityk (in common with other singe purpose mathematical tools) is capable, flexible, and easy to use in isolation. The rub is that, however technically capable such isolated tools may be, any need for multiple interfaces for different purposes means that they are in no way a substitute for an integrated tool - provided that the integrated tool is adequate for the requirement. (Please fixyt). - Please do not reply to this automatically generated notification from Issue Tracker. Please log onto the website and enter your comments. http://qa.openoffice.org/issue_handling/project_issues.html#notification - To unsubscribe, e-mail: issues-unsubscr...@graphics.openoffice.org For additional commands, e-mail: issues-h...@graphics.openoffice.org - To unsubscribe, e-mail: allbugs-unsubscr...@openoffice.org For additional commands, e-mail: allbugs-h...@openoffice.org
[graphics-issues] [Issue 20819] add polynomial regression type
To comment on the following update, log in, then open the issue: http://www.openoffice.org/issues/show_bug.cgi?id=20819 --- Additional comments from hedleyro...@openoffice.org Sun Sep 13 11:39:51 + 2009 --- I'm puzzled by the comment about the problem with the file format. Is the problem in extracting the data, or saving it? In either case, what is the difference between the requirements for polynomials and those for linear regression? While on the subject of the existing fits, although exponential and logarithmic fits are useful in principle, they are only of use if the expression fits both constants (e.g c*e^(a*x). It's also often necessary to use the fitted equation: - it's probably possible at the moment, and I've failed to find it. - Please do not reply to this automatically generated notification from Issue Tracker. Please log onto the website and enter your comments. http://qa.openoffice.org/issue_handling/project_issues.html#notification - To unsubscribe, e-mail: issues-unsubscr...@graphics.openoffice.org For additional commands, e-mail: issues-h...@graphics.openoffice.org - To unsubscribe, e-mail: allbugs-unsubscr...@openoffice.org For additional commands, e-mail: allbugs-h...@openoffice.org
[graphics-issues] [Issue 20819] add polynomial regression type
To comment on the following update, log in, then open the issue: http://www.openoffice.org/issues/show_bug.cgi?id=20819 --- Additional comments from hedleyro...@openoffice.org Sun Sep 13 14:15:11 + 2009 --- hephooey wrote: polynomial regression you need more than 2, and you cannot determine the maximum parameters you need theoretically Maybe I'm misunderstanding what you wrote? My understanding is that generally the user defines the order of fit that is required, so the number of parameters should be fixed at that point. Maybe this is the reason that EXCEL (*$£%) only allows orders up to 6? BTW, my personal preference is an orthogonally-generated Chebyshev fit (see for example http://mathworld.wolfram.com/ChebyshevPolynomialoftheFirstKind.html equation 24). This has the potential advantage that adding an order would not change the values in previous orders, so these can be generated one-at-a-time if required (i.e. only one output variable per order). - Please do not reply to this automatically generated notification from Issue Tracker. Please log onto the website and enter your comments. http://qa.openoffice.org/issue_handling/project_issues.html#notification - To unsubscribe, e-mail: issues-unsubscr...@graphics.openoffice.org For additional commands, e-mail: issues-h...@graphics.openoffice.org - To unsubscribe, e-mail: allbugs-unsubscr...@openoffice.org For additional commands, e-mail: allbugs-h...@openoffice.org
[graphics-issues] [Issue 20819] add polynomial regression type
To comment on the following update, log in, then open the issue: http://www.openoffice.org/issues/show_bug.cgi?id=20819 --- Additional comments from hedleyro...@openoffice.org Sat Aug 15 11:02:25 + 2009 --- I absolutely need polynomial fits for my engineering analyses, most recently to display device repeatability during temperature cycling. For this case, I need access to the polynomial, as I subtracted the polynomial from the data to produce a graph of the cyclic errors. It's been suggested I could write a programme to create polynomial fits - and in principle I could. But if it really is a job that can be done by a user in a sensible time-scale, why is it taking so long to find its way into OpenOffice? So, in spite of everything, it's back to EXCEL. BTW, I'd really like to be able to use minimax algorithms instead of regression. It would also be good to be able to choose the weighting that I apply to the fit (for example data density versus uniform along the line). But the regression technique used by EXCEL does the job (with only a modicum of poet-adjustment), so I'll happily settle for that. - Please do not reply to this automatically generated notification from Issue Tracker. Please log onto the website and enter your comments. http://qa.openoffice.org/issue_handling/project_issues.html#notification - To unsubscribe, e-mail: issues-unsubscr...@graphics.openoffice.org For additional commands, e-mail: issues-h...@graphics.openoffice.org - To unsubscribe, e-mail: allbugs-unsubscr...@openoffice.org For additional commands, e-mail: allbugs-h...@openoffice.org