similar to: How can I calculate the error of a fit parameter, when the data set has an error itself?

Hi out there, imagine you have a dataset (x,y) with errors f, so that each y_i is y_i +- f_i. This is the normal case for almost all measurements, since one quantity y can only be measured with a certain accuracy f. > x<-c(1,2,3) > y<-c(1.1,0.8,1.3) > f<-c(0.2,0.2,0.2) > plot(x,y) #whereas every y has the uncertainty of f If I now perform a nls-fit (and force the data

Dear all, Given a LME model (following the notation of Pinheiro and Bates 2000) y_i = X_i*beta + Z_i*b_i + e_i, is it possible to extract the variance-covariance matrix for the estimated beta_i hat and b_i hat from the lme fitted object? The reason for needing this is because I want to have interval prediction on the predicted values (at level = 0:1). The "predict.lme" seems to

Spencer Thanks for your thoughts on this. I did a bit of work and did end up with a method (more a trick), but it did work. I am certain there are better ways to do this, but here is how I resolved the issue. The integral I need to evaluate is \begin{equation} \frac{\int_c^{\infty} p(x|\theta)f(\theta)d\theta} {\int_{-\infty}^{\infty} p(x|\theta)f(\theta)d\theta} \end{equation} Where

Dear R-list members, I've had a hard time trying to solve a non-linear system (nls) of equations which structure for the equation i, i=1,...,4, is as follows: f_i(d_1,d_2,d_3,d_4)-k_i(l,m,s) = 0 (1) In the expression above, both f_i and k_i are known functions and l, m and s are known constants. I would like to estimate the vector d=(d_1,d_2,d_3,d_4) which is solution

Hi, I am solving a projection pursuit regression problem, of the form y = \sum_i f_i (a_i^T x), where a_i are unknown directions, while f_i are unknown univariate link functions. The following is known about each f_i: 1. f_i (0) = 0  (that is, each f_i passes through the origin) 2. f_i is monotonic. Is there a way to ensure that the function ppr() in R produces solutions that respect the

Hi there Sorry for what may be a naive or dumb question. I have the following data: > x <- c(1,2,3,4) # predictor vector > y <- c(2,4,6,8) # response vector. Notice that it is an exact, perfect straight line through the origin and slope equal to 2 > error <- c(0.3,0.3,0.3,0.3) # I have (equal) ``errors'', for instance, in the measured responses Of course the

Hi, I have a situation where I have a set of pairs of X & Y variables for each of which I have a (fairly) well-defined PDF. The PDF(x_i) 's and PDF(y_i)'s are unfortunately often rather non-Gaussian although most of the time not multi--modal. For these data (estimates of gas content in galaxies), I need to quantify a linear functional relationship and I am trying to do this as

I tried lrm in library(Design) but there is always some error message. Is this function really doing the weighted logistic regression as maximizing the following likelihood: \sum w_i*(y_i*\beta*x_i-log(1+exp(\beta*x_i))) Does anybody know a better way to fit this kind of model in R? FYI: one example of getting error message is like: > x=runif(10,0,3) > y=c(rep(0,5),rep(1,5)) >

> 3x3 subset used > Locus1 Locus2 Locus3 > Samp1 GG <NA> GG > Samp2 AG CA GA > Samp3 AG CA GG > > The euclidean distance function is defined as: sqrt(sum((x_i - y_i)^2)) My > assumption was that the difference between

Hello, I want to solve a nonlinear 3SLS problem with "nlsystemfit()". The equations are of the form y_it = f_i(x,t,theta) The functions f_i(.) have to be formulated as R-functions. When invoking "nlsystemfit()" I get the error Error in deriv.formula(eqns[[i]], names(parmnames)) : Function 'f1' is not in the derivatives table

Dear colleagues, I'm trying (and failing) to write the script required to generate a chart that would help me assess the forecasting accuracy of a logistic regression model by plotting the cumulative proportion of observed events occurring in cases across the range of possible predicted probabilities. In other words, let: x = any value on 0-1 scale phat_i = predicted probability of event Y

I am trying to test out several mgcv::gam models in a scalar-on-function regression analysis. The following is the 'hierarchy' of models I would like to test: (1) Y_i = a + integral[ X_i(t)*Beta(t) dt ] (2) Y_i = a + integral[ F{X_i(t)}*Beta(t) dt ] (3) Y_i = a + integral[ F{X_i(t),t} dt ] equivalents for discrete data might be: 1) Y_i = a + sum_t[ L_t * X_it * Beta_t ] (2) Y_i

This is a covariance calculation question so nothing to do with R but maybe someone could help me anyway. Suppose, I have two random variables X and Y whose means are both known to be zero and I want to get an estimate of their covariance. I have n sample pairs (X1,Y1) (X2,Y2) . . . . . (Xn,Yn) , so that the covariance estimate is clearly 1/n *(sum from i = 1 to n of ( X_i*Y_i) ) But,

Hi the list, According to what I know, the Canberra distance between X et Y is : sum[ (|x_i - y_i|) / (|x_i|+|y_i|) ] (with | | denoting the function 'absolute value') In the source code of the canberra distance in the file distance.c, we find : sum = fabs(x[i1] + x[i2]); diff = fabs(x[i1] - x[i2]); dev = diff/sum; which correspond to the formula : sum[ (|x_i - y_i|) /

Hi, I have a dataset with the following properties: Y_i ~ N(mu_i, theta * (mu_i)^2) ln(mu_i) = B'Xi theta and beta's are the parameters here. I want to come up with a model to fit the data with the above property and test that model on the built in R dataset quine. Does nls() make sense in this case? Or is there any existing R package which can fit this model? -Shelly -- View

for learning purposes and also to help someone, i used roger peng's document to get the mle's of the gamma where the gamma is defined as f(y_i) = (1/gammafunction(shape)) * (scale^shape) * (y_i^(shape-1)) * exp(-scale*y_i) ( i'm defining the scale as lambda rather than 1/lambda. various books define it differently ). i found the likelihood to be n*shape*log(scale) +

Dear R users, I wish to manually demean a panel over time and entities. I tried to code the Wansbeek and Kapteyn (1989) transformation (from Baltagi's book Ch. 9). As a benchmark I use both the pmodel.response() and model.matrix() functions in package plm and the results from using dummy variables. As far as I understood the transformation (Ch.3), Q%*%y (with y being the dependent variable)

Canberra distance is defined in function `dist' (standard library `mva') as sum(|x_i - y_i| / |x_i + y_i|) Obviously this is undefined for cases where both x_i and y_i are zeros. Since double zeros are common in many data sets, this is a nuisance. In our field (from which the distance is coming), it is customary to remove double zeros: contribution to distance is zero when both x_i

Canberra distance is defined in function `dist' (standard library `mva') as sum(|x_i - y_i| / |x_i + y_i|) Obviously this is undefined for cases where both x_i and y_i are zeros. Since double zeros are common in many data sets, this is a nuisance. In our field (from which the distance is coming), it is customary to remove double zeros: contribution to distance is zero when both x_i

Hello, I have the following function that receives a "function pointer" formal parameter name "fnc": loocv <- function(data, fnc) { n <- length(data.x) score <- 0 for (i in 1:n) { x_i <- data.x[-i] y_i <- data.y[-i] yhat <- fnc(x=x_i,y=y_i) score <- score + (y_i - yhat)^2 } score <- score/n