similar to: Which non-parametric regression would allow fitting this type of data? (example given).

Hi, I am trying to normalize some data. First I fitted a principal curve (using the LCPM package), but now I would like to apply a transformation so that the curve becomes a "straight diagonal line" on the plot. The data used to fit the curve would then be normalized by applying the same transformation to it. A simple solution could be to apply translations only (e.g., as done after a

Hi, I posted a message earlier entitled "How to fit a line through the "Mountain crest" ..." I figured loess is probably the best way, but it seems that the problem is the robustness of the fit. Below I paste an example to illustrate the problem: tmp=rnorm(2000) X.background = 5+tmp; Y.background = 5+ (10*tmp+rnorm(2000)) X.specific = 3.5+3*runif(1000);

Sorry, for disturbing the list again. > Also I got one suggestion of using ORDPACK at http://www.netlib.org/. It's ODRPACK at http://www.netlib.org/, not ORDPACK. Best, -- Etsushi Kato ekato at ees.hokudai.ac.jp -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send

Hi, I'm trying to normalize data by fitting a line through the highest density of points (in a 2D plot). In other words, if you visualize the data as a density plot, the fit I'm trying to achieve is the line that goes through the "crest" of the mountain. This is similar yet different to what LOESS does. I've been using loess before, but it does not exactly that as it takes

Hi All: Package TSERIES is stated to depend on MVA. However, there is no MVA package to be found under the list of package sources. Best wishes, ANDREW tseries: Package for time series analysis Package for time series analysis with emphasis on non-linear and non-stationary modelling Version: 0.7-6 Depends: ts, mva, quadprog Date: 2001-08-27 Author: Compiled by Adrian

A student of mine wants to use R to do some nonmetric multidimensional scaling. According to the R FAQ, there's a package called pcurve that computes multidimensional scaling solutions, but I was not able to locate it the contrib page (I am a Windows user with R version 1.4.1). Can anyone tell me whether it is possible to do nonmetric multidimensional scaling with R, and if so, how? John

Dear John, > I suggest that you look at the abilities of the mgcv package. > There are notes of mine at > > http://www.maths.anu.edu.au/%7Ejohnm/r-book/xtras/autosmooth.pdf > > that may help you get started. Thank?you very much for the suggestion and the link to your write-up, it was indeed very helpful! I have experimented with this library for a while now and am really happy

Dear All, I'm sure this is not the first time this question comes up but I couldn't find the keywords that would point me out to it - so apologies if this is a re-post. Basically I've got thousands of points, each depending on three variables: x, y, and z. if I do a plot(x,y, col=z), I get something very messy. So I would like to smooth the values of z according to the values of

Dear R experts, I have a problem that is a related to the question raised in this earlier post https://stat.ethz.ch/pipermail/r-help/2007-January/124064.html My situation is different in that I have only 2 predictors (coordinates x,y) for local regression but a number of global ("parametric") offsets that I need to consider. Essentially, I have a spatial distortion overlaid over a

hi all i have a another stats question. i would like to solve the following question: y(i)=a+b*x(i)+e(i) i.e. estimate a and b (they should be fixed) but i dont want to specify the standard density to the straight line. this can be done using kernel regression. the fitted line is however fitted locally. does anyone have a reference that will help me with my problem. i am still new to

Dear All, I cannot find a solution to the following problem although I imagine that it is a classic, hence my email. I have a vector V of X values comprised between 1 and N. I would like to get random samples of X values also comprised between 1 and N, but the important point is: * I would like to keep the same distribution of distances between the X values * For example let's say N=10 and

Thanks for pointing me out "aggregate", that works fine! There is one complication though: I have mixed types (numerical and character), So the matrix is of the form: A 1.0 200 ID1 A 3.0 800 ID1 A 2.0 200 ID1 B 0.5 20 ID2 B 0.9 50 ID2 C 5.0 70 ID1 One letter always has the same ID but one ID can be shared by many letters (like ID1) I just want to keep track of the ID, and get

Dear R users, I am having trouble devising an efficient way to run a loess() function on all columns of a data.frame (with the x factor remaining the same for all columns) and I was hoping that someone here could help me fix my code so that I won't have to resort to using a for loop. (You'll find the upcoming lines of code in a single block near the end of the message.) Here's a

Hi, Everyone, I want to calculate the principal curve of a points set. First I read the points'coordinate with function "scan", then converted it to matrix with the function "matrix", and fit the curve with function "principal.curve". Here is my data in the file "bmn007.data": 0.023603 -0.086540 -0.001533 0.024349 -0.083877 -0.001454 .. ..

Hey all, I can't seem to get the princurve package to produce correct results, even in the simplest cases. For example, if you just generate a 1 period noiseless sine wave, and ask for the principal curve and plot, the returned curve is clearly wrong (doesn't follow the sine wave). Here's my code: library(princurve) x <- runif(1000,0,2*pi); x <- cbind(x/(2*pi), sin(x)) fit1

Hi, That sounds simple but I cannot think of a really fast way of getting the following: c(1,1,2,2,3,3,4,4) would give c(1,3,5,7) i.e., a function that returns the indexes of the first occurrences of numbers. Note that numbers may have any order e.g., c(3,4,1,2,1,1,2,3,5), can be very large, and the vectors are also very large (which prohibits any loop). The best I could think of is: tmp =

Hi, I've had this problem for a while and tackled it is a quite dirty way so I'm wondering is a better solution exists: If we have two vectors: v1 = c(0,1,2,3,4) v2 = c(5,3,2,1,0) How to remove one instance of the "3,1" / "1,3" double? At the moment I'm using the following solution, which is quite horrible: v1 = c(0,1,2,3,4) v2 = c(5,3,2,1,0) ft <-

Hi, I have a three dimensional array, e.g., my.array = array(0, dim=c(2,3,4), dimnames=list( d1=c("A1","A2"), d2=c("B1","B2","B3"), d3=c("C1","C2","C3","C4")) ) what I would like to get is then a dataframe: d1 d2 d3 value A1 B1 C1 0 A2 B1 C1 0 . . . A2 B3 C4 0 I'm sure there is one function to do

Good evening,I am Marta Colombo, student of "Politecnico di Milano". I'm studying Local Regression Techniques such as loess, smoothing splines and kernel smoothers. Choosing "symmetric" for the argument "family" in loess function it is possible to produce a robust estimate , in function smooth.spline and ksmooth I didn't find this possibility. Well, is there a

Hi all, I am looking for a command for doing 2D filtering (rectangular or Gaussian) in R... I have looked at ksmooth, filter and convolve but they seem to be 1D... Any thoughts? Thanks a lot! [[alternative HTML version deleted]]