similar to: constructing a self-starting non-linear model

Dear all, I would like to fit a non-linear model of the form: y=g*x/(a+b*x) with nls(). However this model is somehow overparameterized and I get the error message about singular gradient matrix at initial parameter estimates. What I am interested in is to make inference about parameters b and g, so this has to be taken into account in the model formulation. What options do I have? Also, how is

Dear all, I have a for loop which includes nls model estimation. The loop breaks after the first non-convergence error. How can I make the loop continue and try to estimate all models? I suppose it should be sth like: if(...) { next } but I have no idea how to setup the arguements... Thank you! Irene

I would like to combine time-series data to test for correlations and interactions using random and fixed effects meta-analysis. So, I am looking for the right packages and documentation. I know about meta and rmeta packages of R. Are there any more? What are the diffrences in brief? Can you please suggest some references that could be used as a guide for meta-analysis in R (or S-plus)?

Hi all! I try to estimate a statistic of the form: (x1-x2)/(y1-y2), where x1,x2,y1,y2 represent variable means, so each has an estimate and standard error associated with it. How is it possible to estimate the mean and the variance of this ratio? Thank you! [[alternative HTML version deleted]]

Hi all! I am trying to run a regression where the predictor values are not real data but each is estimated from a different model. So, for each value I have a mean and variance. Which package/function should I use in this case? Thank you! Irene [[alternative HTML version deleted]]

Hello. I am trying to fit an asymptotic relationship (nonlinear) to some ecological data, and am having problems. I am interested in the upper bound on the data (i.e. if there is an upper limit to 'y' across a range of 'x'). As such, I am using the nonlinear quantile regression package (nlrq) to fit a michaelis mention type model. The errors I get (which are dependant on

Dear list, I would like to produce a plot of variables where the size of the points will be indicative of their standard errors. How is that possible? Thank you! [[alternative HTML version deleted]]

I would like to fit a 2-level mixed model: yit=a+a[i]+a[it] +(b+b[i]+b[it])*xit+eps[it] However, the variance of the second level components should depend on the group, i.e. sigma for a[it] and b[it] should be [i] specific. I do not know whether this is conceptually right in the mixed model context... In case it stands, how should the formula look like? Also, the data are unbalanced with

Hello! I work with : R : Copyright 2006, The R Foundation for Statistical Computing Version 2.3.1 (2006-06-01) On Windows XP Professional (Version 2002) SP2. At this moment I use the function "nls" combined with a selfStar model (SSmicmen, related to Michaelis-Menten equation, and provided by the "stats" package). When I realise the following operation (cf. p 59 of the

I am a new R user and so I thought I could start with "Using R for Introductory statistics" by Verzani. In order to use some of the functions and datasets I have to install the simpleR package which is is now inside the UsingR package. I did so using >install.packages("UsingR"). However, the functions such as "simple.freqpoly.R" do not work. I also tried to

Hello, I'm having problems creating and using a selfStart model with nlme. Briefly, I've defined the model, a selfStart object, and then combined them to make a selfStart.default model. If I apply getInitial to the selfStart model, I get results. However, if I try usint it with nls or nlsList, these routines complain about a lack of initial conditions. If someone could point out

Hi all! How is it possible to estimate standard errors for coef obtained from lme? Is there sth like se.coef() for lmer or what is the anaytical solution? Thank you!

Hi there, I have a data frame as listed below: > Ca.P.Biomass.A P Biomass 1 334.5567 0.2870000 2 737.5400 0.5713333 3 894.5300 0.6393333 4 782.3800 0.5836667 5 857.5900 0.6003333 6 829.2700 0.5883333 I have fit the data using logistic, Michaelis?Menten, and linear model, they all give significance. > fm1 <- nls(Biomass~SSlogis(P, phi1, phi2, phi3), data=Ca.P.Biomass.A)

Dear list, Has anyone created a version of SSmicmen that allows testing for group differences? The basic Michaelis-Menten equation is: (Bmax * X) / (Kd + X). The nlme package allows modeling of random effects for Bmax and Kd as needed, but I curious how I can build in group differences? I have receptor binding data for strains of mice, and following Pinheiro and Bates' lead in their

Hi All, Another nls related problem (for background, I'm migrating a complicated modelling package from S-plus to R). Below I've reduced this to the minimum necessary to demonstrate my problem (I think); the real situation is more complicated. Two similar selfStart functions, ssA and ssB. The 'initial' function for ssB modifies its arguments a little and then calls getInital

Hi, I am a student of Earthquake Engineering, and am new to R. Currently I try to run nonlinear regression analysis by R. My data has three variables: X, Y, and Z. Z is a function of (X, Y). My R script is as below. rm(list=ls()) # read in data Alldata <- read.table("~/Documents/R/Wu_data.dat", header=TRUE) # assign variables Z <- Wu[[1]] # N1,60 X <- Wu[[2]]

Hi all! this is a rather statistical question: is it meaningful to consider an interaction effect between 2 continuous covariates? for example: lm(y~x1+x2+x1:x2) Should one of continuous x1, x2 be "transformed" to a categorical variable, i.e. be classified into groups? Is it easier to interpret the effect if 1 or both are centered to the mean or z-transformed? Thank you!

Dear all, I have an annual time-series of population numbers and I would like to estimate the auto-correlation. Can I use acf() function and judge whether auto-correlation is significant by the plots? The acf array, eg: Autocorrelations of series 'x$log.s.r', by lag 0 1 2 3 4 5 6 7 8 9 10 11 12 1.000 0.031 -0.171

Is there a formal way to prove the need of a mixed model, apart from e.g. comparing the intervals estimated by lmList fit? For example, should I compare (with AIC ML?) a model with seperately (unpooled) estimated fixed slopes (i.e.using an index for each group) with a model that treats this parameter as a random effect (both models treat the remaining parameters as random)? Thank you!

Hi all, I would like to fit a gamm model of the form: Y~X+X*f(z) Where f is the smooth function and With random effects on X and on the intercept. So, I try to write it like this: gam.lme<- gamm(Y~ s(z, by=X) +X, random=list(groups=pdDiag(~1+X)) ) but I get the error message : Error in MEestimate(lmeSt, grps) : Singularity in backsolve at level 0, block 1