OpenOffice.org Forum at OOoForum.orgThe OpenOffice.org Forum
 
 [Home]   [FAQ]   [Search]   [Memberlist]   [Usergroups]   [Register
 [Profile]   [Log in to check your private messages]   [Log in

Regression Line, Curve, Line of best fit etc... on XY Chart

 
Post new topic   Reply to topic    OOoForum.org Forum Index -> OpenOffice.org Calc
View previous topic :: View next topic  
Author Message
Wheelie!
Newbie
Newbie


Joined: 25 Aug 2005
Posts: 4

PostPosted: Thu Aug 25, 2005 2:36 pm    Post subject: Regression Line, Curve, Line of best fit etc... on XY Chart Reply with quote

Hi,

So I have this graph:



And I need to put a line of best fit on it. I have tried all the regression line options, but they either produce a straight horizontal line on the x-axis or a straight line on a gradient, not a curve.

Could anyone tell me how to fix this?

Thanks,
Matt Sad
Back to top
View user's profile Send private message
Dale
Super User
Super User


Joined: 21 Feb 2005
Posts: 1440
Location: Australia

PostPosted: Thu Aug 25, 2005 3:21 pm    Post subject: Reply with quote

Select your data points in the graph
Format > Object Properties
(Or just right-click and select Object Properties)
Select the Statistics tab
Choose the type of regression curve you ant to display.

Also, browse the help index for LOGEST() and LINEST()
_________________
Dale
To err is human, but to destroy your slippers in the process takes a real son of a bitch: Me!

OOo documentation from the source
http://documentation.openoffice.org
Guides, FAQ, How Tos
Back to top
View user's profile Send private message
ftack
Moderator
Moderator


Joined: 27 Jan 2003
Posts: 3102
Location: Belgium

PostPosted: Thu Aug 25, 2005 11:27 pm    Post subject: Reply with quote

There is no build-in options to fit a curve to your datapoints, except for a linear fit which you already found out. You will need to fit the model yourself. Multiple regression can be performed using Calc, but I am not aware of how to easily perform a non-linear regression in Calc. Once you have your model parameters, construct an extra column next to your data to calculate the fitted values and plot them as a line without symbols. I hope you are not using the Beta versions, because charting for me has notably been unstable in all beta's I have seen up to now.
Back to top
View user's profile Send private message
jrkrideau
Super User
Super User


Joined: 08 Aug 2005
Posts: 6732
Location: Kingston ON Canada

PostPosted: Fri Aug 26, 2005 5:17 am    Post subject: Re: Regression Line, Curve, Line of best fit etc... on XY Ch Reply with quote

Wheelie! wrote:
Hi,

So I have this graph:

"Cipped see above post"

And I need to put a line of best fit on it. I have tried all the regression line options, but they either produce a straight horizontal line on the x-axis or a straight line on a gradient, not a curve. Could anyone tell me how to fix this?
Thanks,
Matt Sad


You probably would be much better off with a statistical package that is designed to do this type of work. No spreadsheet is really designed for statistical analysis and curve fitting.

If you have access through your school/work there are some very good commercial packages available: SAS, SYSTAT, Minitab and SPSS are some very good and popular ones and there are many others.

There are free analysis packages available as well:

R is something like the OpenOffice of the statistical world. It is at http://www.r-project.org/. R is very good, not all that easy to learn but very powerful. It has very good regression and graphing abilities and graphs and charts can be imported into OOOo. It is a "sister" package to S. From what I can see the graphics are of publicaton quality and to be honest a lot better than most or all speadsheet graphics, at least for scientific work.

As joerg04 suggested in another thread on regression: “let me advise you to start with the GUI named R Commander or Rcmdr ( http://socserv.mcmaster.ca/jfox/Misc/Rcmdr/ ) which was build by John Fox of the McMaster University (Canada). This will ease the start.”

I followed his advice and he was right. You should be up and running with R plus Rcmdr well enough to do a simple linear regression in less than an hour or so. It may take just a bit longer Smile to figure out how to fit a line to your data.

Note what was not immediately obvious to me was that Rcmdr is not on John Fox's site (the documentation is) but is part of the R package and must be downloaded from the CRAN site just as one would download other packages for the program.

Dataplo, a free package from the US National Institute of Standards and Technology http://www.itl.nist.gov/div898/software/dataplot/ looks like a candidate for what you want but I have not tried it. It is available on most platforms

For other free statistical software have a look at:
http://gsociology.icaap.org/methods/soft.html
http://freestatistics.altervista.org/stat.php
http://www.psychnet-uk.com/experimental_design/software_packages.htm
http://data.fas.harvard.edu/micah_altman/socsci.shtml

.
_________________
jrkrideau
Kingston ON Canada
Currently using Windows 7 & OOo 3.4.0 and Ubuntu 12.04 & LibreOffice 3.5.2.2
Back to top
View user's profile Send private message
enmane
Power User
Power User


Joined: 10 Feb 2004
Posts: 61

PostPosted: Thu Sep 01, 2005 4:06 pm    Post subject: Reply with quote

jrkrideau, I have to disagree with you.


Your excuse isn't a good one. Essentially you are saying - "these packages aren't suited for these functions so don't expect them - go use something else"

There are enough of us that use Excel for these functions and yes, they aren't 100% but if it is good enough then it is good enough. The response that I'd like to see from the OO Calc developers is more like " yes, we realize that our users want this and we are working on bringing it to you even better than what Excel has to offer" instead we get crappy excuses and get pushed away. OOCalc sucks and the people that are watching the development aren't listening to what we want.

I'm pretty darned frustrated with the lack of development in OOCalc.

Instead of recommending another package, how about you OOCalc guys get on the ball and lead people to your package by incorporating these things. It's not rocket science people, it's statistics.
Back to top
View user's profile Send private message
David
Super User
Super User


Joined: 24 Oct 2003
Posts: 5668
Location: Canada

PostPosted: Thu Sep 01, 2005 7:58 pm    Post subject: Reply with quote

enmane wrote:


I'm pretty darned frustrated with the lack of development in OOCalc.

Instead of recommending another package, how about you OOCalc guys get on the ball and lead people to your package by incorporating these things. It's not rocket science people, it's statistics.


We are not all developers. Some are people like yourself, but are those who do not complain when given a great free tool, but rather offer free advice. If it's not rocket science perhaps you might be able to help instead of demanding service? You are getting good advice here, freely offered, about a free program. Mine is that if you want statistical mathematics, then invest in a statistical math program. You might consider Fathom, or more expensive, and more powerful SPSS software. But there are some free ones if you look around, mentioned in this forum from time to time, depending on your needs. If the needs are great, I'd go for the more dedicated statistical software package.

The store I go to, by the way, is the one which serves my needs. If they can send me somewhere else, I thank them for it, then keep coming back.

David.
Back to top
View user's profile Send private message
enmane
Power User
Power User


Joined: 10 Feb 2004
Posts: 61

PostPosted: Thu Sep 01, 2005 8:12 pm    Post subject: Reply with quote

David,
thanks for your wisdom.

I'm not looking a gift horse in the mouth but asking them to improve. If we are to take your advice then no improvement would be found. As someone that has managerial experience I demand that my employees demand more of me so that I can grow. I would hope that there are more like me. I would hope that the OO team would ask for feedback and criticism, no matter how harsh.

That being said, I also think that your advice is misplaced when the OO community is trying to displace MS and promote their package over the competition when it can't do what they do. In this case the saying "either put up or shut up" is applicable. I can hear the presentations now:

Please switch to OO as we are a complete MS Office replacement but you'll need widget A, B, & C to have a complete integrated package that does what MSO does. C'mon people, demand more from the developers. You are cheating them as much as yourself.

I've offered my statistical knowledge to the developers and haven't gotten a lick of a response from any of them. Just look at my bug reports. I'm proactive but my comments are falling on deaf ears.
Back to top
View user's profile Send private message
jrkrideau
Super User
Super User


Joined: 08 Aug 2005
Posts: 6732
Location: Kingston ON Canada

PostPosted: Fri Sep 02, 2005 7:47 am    Post subject: Reply with quote

enmane wrote:
jrkrideau, I have to disagree with you.
Your excuse isn't a good one. Essentially you are saying - "these packages aren't suited for these functions so don't expect them - go use something else"


This is not an excuse, just a statement of fact. Spreadsheets are just not designed to do this type of work well even though they sometimes can do it quite adequately. It's a bit like using a pair of Vise-grips to remove a spark plug. It can work but it may not be the optimal tool for the job. (5)

I have just started using OOo so I am not an expert on graphing in Calc. I do know that there are , more specialized and probably better software packages around. It is up to you to use them or not. I was simply making a suggestion.

Quote:
There are enough of us that use Excel for these functions and yes, they aren't 100% but if it is good enough then it is good enough.

Ye, if it is is good enough then great. The question is "Is it good enough?" I really don't know but I do know that there are available packages out there that are more than good enough for serious plotting and statistical analysis.

Quote:
The response that I'd like to see from the OO Calc developers is more like " yes, we realize that our users want this and we are working on bringing it to you even better than what Excel has to offer" instead we get crappy excuses and get pushed away. OOCalc sucks and the people that are watching the development aren't listening to what we want.

I'm pretty darned frustrated with the lack of development in OOCalc.
Instead of recommending another package, how about you OOCalc guys get on the ball and lead people to your package by incorporating these things. It's not rocket science people, it's statistics.


I am not one of "you OOCalc guys". I'm a new user. I only heard of OOo a month or so ago and so far have used Writer more than anything. Most of my Calc work has been pretty simple-minded. If you want to use a spreadsheet for statistics/ graphing that's fine and often will do what you need. Just be aware that there are better tools out there.

I have not seen any evaluations of Calc for stats and it may well be okay as spreadsheets go. I would appreciate anyone pointing me in the right direction to find such a review. I understand that Gnumeric is quite good. See (4) below. However certainly Excel, as of the latest reports I have read, should not be used for statistics nor, IMO, for most serioius graphing work.

For some comments about Excel and spreadsheets in general when used for stats you might like to look at:
(1) http://www.cs.uiowa.edu/~jcryer/JSMTalk2001.pdf
(2) http://www.practicalstats.com/Pages/excelstats.html
(3) http://pages.stern.nyu.edu/~jsimonof/classes/1305/pdf/excelreg.pdf
(4)http://www.csdassn.org/reportdetail.cfm?ID=508
(5) http://www.burns-stat.com/pages/Tutor/spreadsheet_addiction.html
_________________
jrkrideau
Kingston ON Canada
Currently using Windows 7 & OOo 3.4.0 and Ubuntu 12.04 & LibreOffice 3.5.2.2
Back to top
View user's profile Send private message
David
Super User
Super User


Joined: 24 Oct 2003
Posts: 5668
Location: Canada

PostPosted: Fri Sep 02, 2005 9:34 am    Post subject: Reply with quote

enmane wrote:
David,
thanks for your wisdom.

As someone that has managerial experience I demand that my employees demand more of me so that I can grow. I would hope that there are more like me.


I would sincerely hope that there are not. Frankly, if you were my manager, it would be for a single day. You have received a free program. You have come for free assistance in using it, and all you do is berate those who try to assist you and those who are developing this free program. What do you offer other than that? [The question was rhetorical.]

David.
Back to top
View user's profile Send private message
Display posts from previous:   
Post new topic   Reply to topic    OOoForum.org Forum Index -> OpenOffice.org Calc All times are GMT - 8 Hours
Page 1 of 1

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum


Powered by phpBB © 2001, 2005 phpBB Group