[R-sig-eco] Using multiple species data for gam

Greg Guerin greg.guerin at adelaide.edu.au
Tue Feb 17 00:53:42 CET 2015 

Hello,

I¹m no expert on GLMM convergence issues (and how critical these WARNINGS
are) but there is plenty of information out there or, unless someone else
on this list has some advice, you could try the mixed models list.

This (and links inside) may be a good starting point (concerns very
similar warning messages on an lmer):
http://stats.stackexchange.com/questions/97929/lmer-model-fails-to-converge

Regards,
Greg

--
Dr Greg Guerin
Postdoctoral Fellow
School of Earth and Environmental Sciences, Faculty of Science
The University of Adelaide
Level 12, Schulz Building, North Terrace Campus
greg.guerin at adelaide.edu.au
http://www.adelaide.edu.au/directory/greg.guerin

CRICOS Provider Number 00123M
-----------------------------------------------------------
IMPORTANT: This message may contain confidential or legally privileged
information. If you think it was sent to
you by mistake, please delete all copies and advise the sender. For the
purposes
of the SPAM Act 2003, this
email is authorised by The University of Adelaide.




>
>
>
>----------------------------------------------------------------------
>
>Message: 1
>Date: Mon, 16 Feb 2015 12:28:28 +0530
>From: Rajendra Mohan panda <rmp.iit.kgp at gmail.com>
>Cc: "r-sig-ecology at r-project.org" <r-sig-ecology at r-project.org>
>Subject: Re: [R-sig-eco] Using multiple species data for gam
>Message-ID:
>	<CAGTzHJv7GzSd6UjQH3Oa9Xn+PykLUAbrcZgzFLd=gocGbsudGQ at mail.gmail.com>
>Content-Type: text/plain; charset="UTF-8"
>
>Dear Dr Greg Guerin
>
>Lot many thanks for the advise. I have tried the code (not with 1000sp)
>successfully but I find some warning messages which I need some help
>regarding the same. Mentioned here for your kind advise:
>
> glmer(Response ~ Temp + Pptn + Moisture+Soil+Slope+Aspect+Altitude+(1 +
>Temp + Pptn+
>Moisture+Soil+Slope+Aspect+Altitude|Species),family=binomial(link="logit")
>,
>data = SP)
>Generalized linear mixed model fit by maximum likelihood (Laplace
>Approximation) ['glmerMod']
> Family: binomial  ( logit )
>Formula: Response ~ Temp + Pptn + Moisture + Soil + Slope + Aspect +
>Altitude +
>    (1 + Temp + Pptn + Moisture + Soil + Slope + Aspect + Altitude |
>   Species)
>   Data: SP
>      AIC       BIC    logLik  deviance  df.resid
> 470.1219  748.2537 -191.0610  382.1219      4066
>Random effects:
> Groups  Name        Std.Dev. Corr
> Species (Intercept) 1.19992
>         Temp        0.77083   0.27
>         Pptn        0.07742  -0.91 -0.64
>         Moisture    0.30603  -0.50 -0.91  0.79
>         Soil        1.17142  -0.44 -0.93  0.75  0.99
>         Slope       0.67329  -0.39 -0.53  0.56  0.32  0.31
>         Aspect      0.21153  -0.90 -0.65  1.00  0.82  0.78  0.50
>         Altitude    0.24942  -0.92 -0.63  1.00  0.78  0.74  0.55  1.00
>Number of obs: 4110, groups:  Species, 5
>Fixed Effects:
>(Intercept)         Temp         Pptn     Moisture         Soil
> Slope       Aspect     Altitude
> -8.8215831    0.3879856    0.0002847   -0.0816050   -0.1431987
>-0.6200528    0.0116070   -0.0003179
>Warning messages:
>1: Some predictor variables are on very different scales: consider
>rescaling
>2: In commonArgs(par, fn, control, environment()) :
>  maxfun < 10 * length(par)^2 is not recommended.
>3: In (function (fn, par, lower = rep.int(-Inf, n), upper = rep.int(Inf,
>:
>  failure to converge in 10000 evaluations
>4: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv,  :
>  Model failed to converge with max|grad| = 37.0466 (tol = 0.001,
>component
>1)
>5: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv,  :
>  Model failed to converge: degenerate  Hessian with 7 negative
>eigenvalues
>
>When I run nlmer, I find the following error message for which I also need
>some help also because my data are mostly non-linear.
>
>Your guidance will be highly appreciated
>
>
>
>With Best Regards
>Rajendra M Panda
>SWR, IIT Kharagpur, India
>
>On Thu, Feb 12, 2015 at 5:08 AM, Greg Guerin <greg.guerin at adelaide.edu.au>
>wrote:
>
>> Hello,
>>
>> not sure if you are looking to run the GLM/GAMs individually but in one
>> run, or as a community composition type model to test main
>> drivers/correlates of combined species occurrences. If the latter,
>>another
>> option is a GLMM with species having random slope to allow responses to
>> differ. For this, you would need to stack the occurrence matrix into a
>> ?long? format (a row for the presence/absence of each species in each
>>plot
>> with corresponding predictor variables and a field for species).
>>
>> Response Species Temp Pptn
>>
>> 0       Sp1     30      1000
>> 1       Sp2     30      1000
>> 1       Sp3     30      1000
>>
>> In lme4, something like:
>> lmer(Response ~ Temp + Pptn + (1 + Temp + Pptn|Species),
>> family=binomial(link="logit"), data)
>>
>> An example with R code in the Appendix:
>> http://dx.doi.org/10.1111/jvs.12111
>>
>> Greg
>>
>> --
>> Dr Greg Guerin
>> Postdoctoral Fellow
>> School of Biological Sciences, Faculty of Science
>> The University of Adelaide
>>
>> CRICOS Provider Number 00123M
>> -----------------------------------------------------------
>> IMPORTANT: This message may contain confidential or legally privileged
>> information. If you think it was sent to
>> you by mistake, please delete all copies and advise the sender. For the
>> purposes
>> of the SPAM Act 2003, this
>> email is authorised by The University of Adelaide.
>>
>>
>>
>>
>>
>> >
>> >----------------------------------------------------------------------
>> >
>> >Message: 1
>> >Date: Tue, 10 Feb 2015 09:28:14 -0700
>> >From: Tim Meehan <tmeeha at gmail.com>
>> >To: Rajendra Mohan panda <rmp.iit.kgp at gmail.com>
>> >Cc: "r-sig-ecology at r-project.org" <r-sig-ecology at r-project.org>
>> >Subject: Re: [R-sig-eco] Using multiple species data for gam
>> >Message-ID:
>> >       <
>> CAMTWOzpv58RRX2ocgTCpXh1EPAcxzkSCGKHvdUsaytGhCJH8MQ at mail.gmail.com>
>> >Content-Type: text/plain; charset="UTF-8"
>> >
>> >If you want to do this in a glm framework, you might look into the
>>mvabund
>> >package:
>> >
>> >http://cran.r-project.org/web/packages/mvabund/mvabund.pdf
>> >
>> >I've never used it with anything approaching 1000 species, though.
>> >
>> >On Tue, Feb 10, 2015 at 2:41 AM, Rajendra Mohan panda
>> ><rmp.iit.kgp at gmail.com
>> >> wrote:
>> >
>> >> Dear All
>> >>
>> >> I have >1000 species with presence and absence (0 or 1) values and
>>with
>> >> seven corresponding predictor variables. If I can run gam/glm for the
>> >>data
>> >> using all species data simultaneously vs predictors. Data are
>>arranged
>> >>in
>> >> columns against their GPS locations (see below). I know it is
>>possible
>> >>to
>> >> do separately for each species.
>> >>
>> >> Your kind response is highly appreciated.
>> >>
>> >> Sites  Sp1  Sp2 Sp3 Alt Temp Pptn   Ft
>> >> 1A         0      1    1     20   30     1000 Evergreen
>> >>
>> >> With Best Regards
>> >> Rajendra M Panda
>> >> School of Water Resources
>> >> Indian Institute of Technology Kharagpur, India
>> >>
>> >>         [[alternative HTML version deleted]]
>> >>
>> >> _______________________________________________
>> >> R-sig-ecology mailing list
>> >> R-sig-ecology at r-project.org
>> >> https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
>> >>
>> >
>> >       [[alternative HTML version deleted]]
>> >
>> >

More information about the R-sig-ecology mailing list