thr3ads.net - R help - [R] Singularity in a regression? [Feb 2009]

If this information is useful, please help other people find it:
Share via:

Bob Gotwals

2009-Feb-26 12:58 UTC

[R] Singularity in a regression?

R friends,

In a matrix of 1s and 0s, I'm getting a singularity error.  Any helpful
ideas?

lm(formula = activity ~ metaF + metaCl + metaBr + metaI + metaMe +
     paraF + paraCl + paraBr + paraI + paraMe)

Residuals:
        Min         1Q     Median         3Q        Max
-4.573e-01 -7.884e-02  3.469e-17  6.616e-02  2.427e-01

Coefficients: (1 not defined because of singularities)
             Estimate Std. Error t value Pr(>|t|)
(Intercept)   7.9173     0.1129  70.135  < 2e-16 ***
metaF        -0.3973     0.2339  -1.698 0.115172
metaCl            NA         NA      NA       NA
metaBr        0.3454     0.1149   3.007 0.010929 *
metaI         0.4827     0.2339   2.063 0.061404 .
metaMe        0.3654     0.1149   3.181 0.007909 **
paraF         0.7675     0.1449   5.298 0.000189 ***
paraCl        0.3400     0.1449   2.347 0.036925 *
paraBr        1.0200     0.1449   7.040 1.36e-05 ***
paraI         1.3327     0.2339   5.697 9.96e-05 ***
paraMe        1.2191     0.1573   7.751 5.19e-06 ***
---
Signif. codes:  0 ?***? 0.001 ?**? 0.01 ?*? 0.05 ?.? 0.1 ? ? 1

Residual standard error: 0.2049 on 12 degrees of freedom
Multiple R-squared: 0.9257,	Adjusted R-squared: 0.8699
F-statistic: 16.61 on 9 and 12 DF,  p-value: 1.811e-05

(Ted Harding)

2009-Feb-26 15:07 UTC

head link

[R] Singularity in a regression?

On 26-Feb-09 12:58:49, Bob Gotwals wrote:> R friends,
> 
> In a matrix of 1s and 0s, I'm getting a singularity error.  Any helpful
> ideas?
>From the degress of freedom in your output, it seems you are fitting10 binary variables to a total of 23 observations. In such circumstances,
it is not unlikely that the matrix of 0s and 1s representing the
binary variables would have at least 1 column which can be represented
as a linear combination of the others (which is what the "1 not defined
because of singularities" means). Get more data, or use fewer variables!
Or, also worth considering, check whether there are relationahips
"in the real world" between your 10 variables which would tend to
generate such linear dependence.

Ted.
> lm(formula = activity ~ metaF + metaCl + metaBr + metaI + metaMe +
>      paraF + paraCl + paraBr + paraI + paraMe)
> 
> Residuals:
>         Min         1Q     Median         3Q        Max
> -4.573e-01 -7.884e-02  3.469e-17  6.616e-02  2.427e-01
> 
> Coefficients: (1 not defined because of singularities)
>              Estimate Std. Error t value Pr(>|t|)
> (Intercept)   7.9173     0.1129  70.135  < 2e-16 ***
> metaF        -0.3973     0.2339  -1.698 0.115172
> metaCl            NA         NA      NA       NA
> metaBr        0.3454     0.1149   3.007 0.010929 *
> metaI         0.4827     0.2339   2.063 0.061404 .
> metaMe        0.3654     0.1149   3.181 0.007909 **
> paraF         0.7675     0.1449   5.298 0.000189 ***
> paraCl        0.3400     0.1449   2.347 0.036925 *
> paraBr        1.0200     0.1449   7.040 1.36e-05 ***
> paraI         1.3327     0.2339   5.697 9.96e-05 ***
> paraMe        1.2191     0.1573   7.751 5.19e-06 ***
> ---
> Signif. codes:  0 ?***? 0.001 ?**? 0.01 ?*? 0.05 ?.? 0.1 ? ? 1
> 
> Residual standard error: 0.2049 on 12 degrees of freedom
> Multiple R-squared: 0.9257,   Adjusted R-squared: 0.8699
> F-statistic: 16.61 on 9 and 12 DF,  p-value: 1.811e-05
> 
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
--------------------------------------------------------------------
E-Mail: (Ted Harding) <Ted.Harding at manchester.ac.uk>
Fax-to-email: +44 (0)870 094 0861
Date: 26-Feb-09                                       Time: 15:07:40
------------------------------ XFMail ------------------------------

ONKELINX, Thierry

2009-Feb-26 15:12 UTC

head link

[R] Singularity in a regression?

It looks like your data has not enough information to estimate the
parameter for metaCl. Maybe because metaCL is identical to one of the
other variables or a constant.
 
HTH,

Thierry

------------------------------------------------------------------------
----
ir. Thierry Onkelinx
Instituut voor natuur- en bosonderzoek / Research Institute for Nature
and Forest
Cel biometrie, methodologie en kwaliteitszorg / Section biometrics,
methodology and quality assurance
Gaverstraat 4
9500 Geraardsbergen
Belgium 
tel. + 32 54/436 185
Thierry.Onkelinx at inbo.be 
www.inbo.be 

To call in the statistician after the experiment is done may be no more
than asking him to perform a post-mortem examination: he may be able to
say what the experiment died of.
~ Sir Ronald Aylmer Fisher

The plural of anecdote is not data.
~ Roger Brinner

The combination of some data and an aching desire for an answer does not
ensure that a reasonable answer can be extracted from a given body of
data.
~ John Tukey

-----Oorspronkelijk bericht-----
Van: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org]
Namens Bob Gotwals
Verzonden: donderdag 26 februari 2009 13:59
Aan: r-help at r-project.org
Onderwerp: [R] Singularity in a regression?

R friends,

In a matrix of 1s and 0s, I'm getting a singularity error.  Any helpful
ideas?

lm(formula = activity ~ metaF + metaCl + metaBr + metaI + metaMe +
     paraF + paraCl + paraBr + paraI + paraMe)

Residuals:
        Min         1Q     Median         3Q        Max
-4.573e-01 -7.884e-02  3.469e-17  6.616e-02  2.427e-01

Coefficients: (1 not defined because of singularities)
             Estimate Std. Error t value Pr(>|t|)
(Intercept)   7.9173     0.1129  70.135  < 2e-16 ***
metaF        -0.3973     0.2339  -1.698 0.115172
metaCl            NA         NA      NA       NA
metaBr        0.3454     0.1149   3.007 0.010929 *
metaI         0.4827     0.2339   2.063 0.061404 .
metaMe        0.3654     0.1149   3.181 0.007909 **
paraF         0.7675     0.1449   5.298 0.000189 ***
paraCl        0.3400     0.1449   2.347 0.036925 *
paraBr        1.0200     0.1449   7.040 1.36e-05 ***
paraI         1.3327     0.2339   5.697 9.96e-05 ***
paraMe        1.2191     0.1573   7.751 5.19e-06 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05
'.' 0.1 ' ' 1

Residual standard error: 0.2049 on 12 degrees of freedom
Multiple R-squared: 0.9257,	Adjusted R-squared: 0.8699
F-statistic: 16.61 on 9 and 12 DF,  p-value: 1.811e-05

______________________________________________
R-help at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Dit bericht en eventuele bijlagen geven enkel de visie van de schrijver weer 
en binden het INBO onder geen enkel beding, zolang dit bericht niet bevestigd is
door een geldig ondertekend document. The views expressed in  this message 
and any annex are purely those of the writer and may not be regarded as stating 
an official position of INBO, as long as the message is not confirmed by a duly 
signed document.

David Winsemius

2009-Feb-26 15:25 UTC

head link

[R] Singularity in a regression?

I saw Ted's reply and it is certainly sensible. I would wonder whether  
to model ought to be recast so that the scientific question is more  
clear? You are obviously studying the effect of different  
substitutions (F, Cl, Br, I, Me) and different positions around an  
aromatic ring (meta, para). Why not consider the order of  
electrophilicity (or possibly size) and the position as two different  
variables, one ordered and the other binomial?

After recoding, your formula might then look like  activity ~ electro  
+ position ... or possibly activity ~ electro + size + position, and  
you would be less likely to run into difficulties with collinearity.  
You would also have some science in your model rather than casting  
aimlessly about in the data. If your ordering is sensible, you end up  
testing with 2 or 3 degrees of freedom.

-- 
David Winsemius

On Feb 26, 2009, at 7:58 AM, Bob Gotwals wrote:
> R friends,
>
> In a matrix of 1s and 0s, I'm getting a singularity error.  Any  
> helpful ideas?
>
> lm(formula = activity ~ metaF + metaCl + metaBr + metaI + metaMe +
>    paraF + paraCl + paraBr + paraI + paraMe)
>
> Residuals:
>       Min         1Q     Median         3Q        Max
> -4.573e-01 -7.884e-02  3.469e-17  6.616e-02  2.427e-01
>
> Coefficients: (1 not defined because of singularities)
>            Estimate Std. Error t value Pr(>|t|)
> (Intercept)   7.9173     0.1129  70.135  < 2e-16 ***
> metaF        -0.3973     0.2339  -1.698 0.115172
> metaCl            NA         NA      NA       NA
> metaBr        0.3454     0.1149   3.007 0.010929 *
> metaI         0.4827     0.2339   2.063 0.061404 .
> metaMe        0.3654     0.1149   3.181 0.007909 **
> paraF         0.7675     0.1449   5.298 0.000189 ***
> paraCl        0.3400     0.1449   2.347 0.036925 *
> paraBr        1.0200     0.1449   7.040 1.36e-05 ***
> paraI         1.3327     0.2339   5.697 9.96e-05 ***
> paraMe        1.2191     0.1573   7.751 5.19e-06 ***
> ---
> Signif. codes:  0 ?***? 0.001 ?**? 0.01 ?*? 0.05 ?.? 0.1 ? ? 1
>
> Residual standard error: 0.2049 on 12 degrees of freedom
> Multiple R-squared: 0.9257,	Adjusted R-squared: 0.8699
> F-statistic: 16.61 on 9 and 12 DF,  p-value: 1.811e-05
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.

Alex Roy

2009-Feb-27 14:07 UTC

head link

[R] Singularity in a regression?

If  collinearity exists, one of the solutions is regulazation version of
regression. There are different types of regularization method. like Ridge,
LASSO, elastic net etc. For example, in  MASS package you can get ridge
regression.

Alex


On Thu, Feb 26, 2009 at 1:58 PM, Bob Gotwals <gotwals@ncssm.edu> wrote:
> R friends,
>
> In a matrix of 1s and 0s, I'm getting a singularity error.  Any helpful
> ideas?
>
> lm(formula = activity ~ metaF + metaCl + metaBr + metaI + metaMe +
>    paraF + paraCl + paraBr + paraI + paraMe)
>
> Residuals:
>       Min         1Q     Median         3Q        Max
> -4.573e-01 -7.884e-02  3.469e-17  6.616e-02  2.427e-01
>
> Coefficients: (1 not defined because of singularities)
>            Estimate Std. Error t value Pr(>|t|)
> (Intercept)   7.9173     0.1129  70.135  < 2e-16 ***
> metaF        -0.3973     0.2339  -1.698 0.115172
> metaCl            NA         NA      NA       NA
> metaBr        0.3454     0.1149   3.007 0.010929 *
> metaI         0.4827     0.2339   2.063 0.061404 .
> metaMe        0.3654     0.1149   3.181 0.007909 **
> paraF         0.7675     0.1449   5.298 0.000189 ***
> paraCl        0.3400     0.1449   2.347 0.036925 *
> paraBr        1.0200     0.1449   7.040 1.36e-05 ***
> paraI         1.3327     0.2339   5.697 9.96e-05 ***
> paraMe        1.2191     0.1573   7.751 5.19e-06 ***
> ---
> Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
>
> Residual standard error: 0.2049 on 12 degrees of freedom
> Multiple R-squared: 0.9257,     Adjusted R-squared: 0.8699
> F-statistic: 16.61 on 9 and 12 DF,  p-value: 1.811e-05
>
> ______________________________________________
> R-help@r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
>
http://www.R-project.org/posting-guide.html<http://www.r-project.org/posting-guide.html>
> and provide commented, minimal, self-contained, reproducible code.
>
	[[alternative HTML version deleted]]

Reasonably Related Threads

Search for more apparently analagous threads

R help - Feb 2009 - Singularity in a regression?

[R] Singularity in a regression?

[R] Singularity in a regression?

[R] Singularity in a regression?

[R] Singularity in a regression?

[R] Singularity in a regression?

Reasonably Related Threads