[R] Fwd: Re: lmer causes R session to terminate

[Bert Gunter]

I failed to c.c. the list. And it should be r-sig-mixed-models list.

Bert
---------- Forwarded message ----------
From: "Bert Gunter" <bgunter.4567 at gmail.com>
Date: Jul 7, 2016 2:06 PM
Subject: Re: [R] lmer causes R session to terminate
To: "David Kikuchi" <dwkikuchi at gmail.com>
Cc:

Off the top, I would guess that you'd need 10 or more times the data to get
> meaningful estimates of the variance components, and heaven knows what
> sorts of rank deficient matrices imbalance in the data may be producing.
> But post on the r-mixed-models list to  get an authoritative answer, which
> this ain't.
>
> Cheers,
> Bert
> On Jul 7, 2016 10:55 AM, "David Kikuchi" <dwkikuchi at gmail.com> wrote:
>
>> Hi,
>>
>> I am working with a large dataset of neotropical birds, and am trying to
>> partition the variance in log(body mass) within different taxonomic
>> levels.
>> To better explain what I mean, the taxonomic levels are species, genus,
>> family, and order. Species are within genera, genera are within famillies,
>> and famlies are within orders.
>>
>> Sample data look like this:
>>
>> mass        species                   genus         family         order.2
>> 377.0000    Geranospiza caerulescens  Geranospiza   Accipitridae
>> Accipitriformes
>> 213.1667    Harpagus bidentatus       Harpagus      Accipitridae
>> Accipitriformes
>> 500.0000    Leptodon cayanensis       Leptodon      Accipitridae
>> Accipitriformes
>> 1750.0000   Penelope albipennis       Penelope      Cracidae
>>  Galliformes
>> 278.0000    Leucopternis semiplumbeus Leucopternis  Accipitridae
>> Accipitriformes
>> 66.2500     Notharchus pectoralis     Notharchus    Bucconidae
>> Galibuliformes
>> 213.1667    Harpagus bidentatus       Harpagus      Accipitridae
>> Accipitriformes
>> 31.0000     Gymnopithys leucaspis     Gymnopithys   Thamnophilidae
>> Passeriformes
>> 31.0000     Gymnopithys leucaspis     Gymnopithys   Thamnophilidae
>> Passeriformes
>>
>>
>> I want to know how much variability in log(mass) there is at the species
>> level, genus level, family level, and order level.
>>
>> The code that I have been using to do this looks like:
>>
>> bm.full <- lmer(log(mass) ~ 1 + (1|order.2/family/genus/species))
>>
>> however, it crashes R every time, whether I run it on my laptop with 4 gb
>> RAM or a desktop with 8 gb RAM. If I remove any of the taxonomic levels,
>> though, it does run and produces reasonable output on either machine.
>> There
>> are 1943 observations (different studies occassionaly give different
>> masses
>> for the same species), 791 species, 381 genera, 60 families, and 21
>> orders.
>> If I am able to modestly increase RAM, e.g. to 16 GB, is it likely that R
>> will be able to handle the model, or is there such a dramatic increase in
>> the computation required with four nested groups that it simply won't be
>> possible?
>>
>> Thank you for your advice,
>>
>> David
>>
>>         [[alternative HTML version deleted]]
>>
>> ______________________________________________
>> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
>> 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.
>>
>

	[[alternative HTML version deleted]]