R help - Feb 2011 - Error in solve.default(inf, tol = tol.solve) :

Hello,



I'm trying to run a lagsarlm (maximum likelihood estimation of a spatial lag
model)  in the spdep library ; but R gives following error message:



Error in solve.default(inf, tol = tol.solve) :

  system is computationally singular: reciprocal condition number = 4.20137e-12



I get the same message when I try to run de lagsarlm with a bigger data set
(4333 regions).



The command I used:

TestLag<-lagsarlm( mean_price ~  transcount+  C1_5_1+ C1_5_2+ C1_5_3+ C1_5_4,
data=DataB,transcecB.listw)



summary(transcecB.listw)

Characteristics of weights list object:

Neighbour list object:

Number of regions: 521

Number of nonzero links: 2904

Percentage nonzero weights: 1.069846

Average number of links: 5.573896

Link number distribution:



  1   2   3   4   5   6   7   8   9  10  11  13

  2  17  46  86 116 108  68  43  23   7   2   3

2 least connected regions:

12070513 12070504 with 1 link

3 most connected regions:

12060503 11040601 11020701 with 13 links



Weights style: W

Weights constants summary:

    n     nn  S0       S1       S2

W 521 271441 521 200.3434 2169.626



If any further information is needed, please let me know.



I hope this is the right platform to get some help in solving this problem.



Best regards,



Marie-Line Glaesener



Phd. Student

Unité de Recherche IPSE (Identités. Politiques, Sociétés, Espaces) Laboratoire
de Géographie et Aménagement du Territoire



UNIVERSITÉ DU LUXEMBOURG

CAMPUS WALFERDANGE

Route de Diekirch / BP 2

L-7201 Walferdange

Luxembourg

www.geo.ipse.uni.lu<http://www.geo.ipse.uni.lu>

	[[alternative HTML version deleted]]

On 04.02.2011 08:37, Marie-Line Glaesener wrote:
> Hello,
>
>
>
> I'm trying to run a lagsarlm (maximum likelihood estimation of a
spatial lag model)  in the spdep library ; but R gives following error message:
>
>
>
> Error in solve.default(inf, tol = tol.solve) :
>
>    system is computationally singular: reciprocal condition number =
4.20137e-12
May we guess that you have almost perfectly correlated columns in your 
data - or factor levels with very few observations? Or many levels? But 
since we cannot know since we do not have your data. Read about how to 
invert matrices numerically and you will find that it is not easy if 
your matrix is almost singular.

Uwe Ligges


>
>
> I get the same message when I try to run de lagsarlm with a bigger data set
(4333 regions).
>
>
>
> The command I used:
>
> TestLag<-lagsarlm( mean_price ~  transcount+  C1_5_1+ C1_5_2+ C1_5_3+
C1_5_4, data=DataB,transcecB.listw)
>
>
>
> summary(transcecB.listw)
>
> Characteristics of weights list object:
>
> Neighbour list object:
>
> Number of regions: 521
>
> Number of nonzero links: 2904
>
> Percentage nonzero weights: 1.069846
>
> Average number of links: 5.573896
>
> Link number distribution:
>
>
>
>    1   2   3   4   5   6   7   8   9  10  11  13
>
>    2  17  46  86 116 108  68  43  23   7   2   3
>
> 2 least connected regions:
>
> 12070513 12070504 with 1 link
>
> 3 most connected regions:
>
> 12060503 11040601 11020701 with 13 links
>
>
>
> Weights style: W
>
> Weights constants summary:
>
>      n     nn  S0       S1       S2
>
> W 521 271441 521 200.3434 2169.626
>
>
>
> If any further information is needed, please let me know.
>
>
>
> I hope this is the right platform to get some help in solving this problem.
>
>
>
> Best regards,
>
>
>
> Marie-Line Glaesener
>
>
>
> Phd. Student
>
> Unit? de Recherche IPSE (Identit?s. Politiques, Soci?t?s, Espaces)
Laboratoire de G?ographie et Am?nagement du Territoire
>
>
>
> UNIVERSIT? DU LUXEMBOURG
>
> CAMPUS WALFERDANGE
>
> Route de Diekirch / BP 2
>
> L-7201 Walferdange
>
> Luxembourg
>
> www.geo.ipse.uni.lu<http://www.geo.ipse.uni.lu>
>
> 	[[alternative HTML version deleted]]
>
>
>
>
> ______________________________________________
> 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.