similar to: Significance of confidence intervals in the Non-Linear Least Squares Program.

Thanks for the response. I was not very clear in my original request. What I am asking is if in a non-linear estimation problem using nls(), as the condition number of the Hessian matrix becomes larger, will the t-values of one or more of the parameters being estimated in general become smaller in absolute value -- that is, are low t-values a sign of an ill-conditioned Hessian? Typical

Hi all, While running the anscombe.test in R, I'm getting an error of *Error in if (pval > 1) pval <- 2 - pval : missing value where TRUE/FALSE needed* for a few time series columns whereas for most of the series the function is working fine. I have checked for those specific columns for missing values. However, there is no NA/NAN value in the dataset. I have also run kurtosis for

Folks: I'm trying to port some code from python over to R, and I'm running into a wall finding R code that can solve a generalized eigenvalue problem following this function model: http://docs.scipy.org/doc/scipy/reference/generated/scipy.linalg.eig.html Any ideas? I don't want to call python from within R for various reasons, I'd prefer a "native" R solution if one

Hi, I have three matrices (X,Y,P) with the same dimension. The X,Y grid is regular and I want to perform linear interpolation to pick out certain points. In matlab appropriate call is something like Pout=interp2(X,Y,P,Xout,Yout, method="linear") where Xout and Yout are the locations where I want the Pout data (typically a different grid). (Scipy has this routine in

HI, I've noticed what I think is an incorrect behavior of stats::psmirnov function and consequently of ks.test when run in an exact mode. For example: psmirnov(1, sizes=c(50, 50), z=1:100, two.sided = FALSE, lower.tail = F, exact=TRUE) produces 2.775558e-15 However, the exact value should be 1/combination(100, 50), which is 9.9e-30. While the absolute error is small, the relative error is

moments::anscombe.test(x) does give errors when x has too few values or if all the values in x are the same > moments::anscombe.test(c(1,2,6)) Error in if (pval > 1) pval <- 2 - pval : missing value where TRUE/FALSE needed > moments::anscombe.test(c(2,2,2,2,2,2,2,2)) Error in if (pval > 1) pval <- 2 - pval : missing value where TRUE/FALSE needed You can use tryCatch() to

Hi Bill, Thanks for prompt reply and letting me know a way around it. I have more than 1200 observations and not all the values are the same. However, my data points are quite similar, for example, 0.079275, 0.078867, 0.070716 in millions and etc. I have run the data without converting it to millions and I still get the same error message. As I have kurtosis value, it should be fine for the

Dear all I am having a linear system of the form A*X=B and I want to find the X by using least squares. For example my A is of dimension [205,3] and my B is of dimension[205,1]  I am looking for the X matrix which is of the size [3,1]. In the matlab I was doing that by the function  X = LSCOV(A,B) returns the ordinary least squares solution to the     linear system of equations A*X = B, i.e., X

Dear All, Which package/function could i use to solve following linear least square problem? A over determined system of linear equations is given. The nnls-function may would be a possibility BUT: The solving is constrained with a inequality that all unknowns are >= 0 and a equality that the sum of all unknowns is 1 The influence of the equations according to the solving process is

I am using "nls" to fit dose-response curves but am not sure how to approach more robust regression in R to get around the problem of the my error showing increased variance with increasing dose. My understanding is that "rlm" or "lqs" would not be a good idea here. 'Fairly new to regression work, so apologies if I'm missing something obvious.

I need to find out the difference between the way R calculates weighted regression and standard regression. I want to plot a 95% confidence interval around an estimte i got from least squares regression. I cant find he documentation for this ive looked in ?stats ?lm ?predict.lm ?weights ?residuals.lm Can anyone shed light? thanks Chris. -- View this message in context:

Dear R-ians, I want to perform an linear unmixing of image pixels in fractions of pure endmembers. Therefore I need to perform a constrained linear least-squares problem that looks like : min || Cx - d || ? where sum(x) = 1. I have a 3x3 matrix C, containing the values for endmembers and I have a 3x1 column vector d (for every pixel in the image). In theory my x values should all be in the

HI, I'd like to perform a weighted linear least squares fit with R on data with varying errors on both vectors. I can do this with one axis using lm, but have no idea where to go from here. I've tried googling, but no idea. Any suggestions? Thanks, James

Hi, everyone, My question is: It's not every time that you can get a converged result from the nls function. Is there any solution for me to get a reasonable result? For example: x <- c(-0.06,-0.04,-0.025,-0.015,-0.005,0.005,0.015,0.025,0.04,0.06) y <- c(1866760,1457870,1314960,1250560,1184850,1144920,1158850,1199910,1263850,1452520) fitOup<- nls(y ~ constant + A*(x-MA)^4 +

Another bad case is > moments::anscombe.test(rep(c(1,1.1),length=35)) Error in if (pval > 1) pval <- 2 - pval : missing value where TRUE/FALSE needed I haven't checked the formulas carefully, but I suspect the problem is from taking the cube root of a negative number in z <- (1 - 2/(9 * a) - ((1 - 2/a)/(1 + xx * sqrt(2/(a - 4))))^(1/3))/sqrt(2/(9 * a)) In R, the

Dear list, I have encountered a special case for searching a linear regression where I'm not satisfied with the results obtained using the traditional least squares method (sometimes called OLS) for estimating/optimizing the residues to the regression line (see code below). Basically, a group of my x-y data are a bit off the diagonal line (in my case the diagonal represents the ideal or

I agree with many the sentiments about the wisdom of computing very small p-values (although the example below may win some kind of a prize: I've seen people talking about p-values of the order of 10^(-2000), but never 10^(-(10^8)) !). That said, there are a several tricks for getting more reasonable sums of very small probabilities. The first is to scale the p-values by dividing the

Per the recent (seeming) consensus regarding incubating new projects under the LLVM organization, I would like to trial the process by requesting to incubate mlir-npcomp <https://github.com/google/mlir-npcomp>. The project is still quite young and has been primarily developed part time by myself and Sean Silva over the last ~2 months. We set it up following discussion of a Numpy/Scipy op set

Dear all, After a thorough research, I still find myself unable to find a function that does linear regression of 2 vectors of data using the "total least squares", also called "orthogonal regression" (see : http://en.wikipedia.org/wiki/Total_least_squares) instead of the "ordinary least squares" method. Indeed, the "lm" function has a

Dear R users, I would like to make my R code for MCMC faster. It is possible to integrate C code into R but I think C is too complicated for me. I would need a C introduction only for MCMC and I do not know if such a thing exists. I was thinking of Python (and scipy). Where could I read about its integration into R ? How developed are the statistical packages in Python ? I could not find a