similar to: Multivariate time series - Poisson with delayed lags

Hi, I've been searching CRAN and the web for a recursive multivariate filter with time-varying coefficients. What I mean is the following: I have a series of square matrices A_t an initial value vector y_0 and I need to compute y_t =A_t%*%y_t-1 As these y_t may diverge quickly and/or lead to underflow problems, the y_t need to be scaled by eg y_t =y_t/sum(y_t-1) Is anyone aware

Dear Professor Bernhard, Sorry for take your time, but I found something strange that I am not able to explain/understand. Suppose that I compute the ADF test by using the criterion="BIC" to select the lags: summary(CADFtest(y, max.lag.y = 20, type = "drift", criterion="BIC")) Suppose that 2 lags are selected. Next, if I set the lags to 2: summary(CADFtest(y,

Dear useRs, I've release a new version of the "dynlm" package to CRAN which adds two new features: o instrumental variables regression (two-stage least squares) via formulas like dynlm(y ~ x1 + x2 | z1 + z2 + z3, data = mydata) where z1, z2, z3 are the instruments which can again contain lags/differences/season via the d()/L()/season() operators. o

hi: you should probably send below to R-Sig-Finance because there are some econometrics people over there who could also possibly give you a good answer and may not see this email ? Also, there's package called mar ( I think that's the name ) that may do what you want ? Finally, I don't know how to do it but I think there are ways of converting a multivariate arima into the

Hi all, I'm new with R (and S), and relatively new to statistics (I'm a computer scientist), so I ask sorry in advance if my question is silly. My problem is this: I have a (sample of a) discrete time stochastic process {X_t} and I want to estimate Pr{ X_t | X_{t-l_1}, X_{t-l_2}, ..., X_{t-l_k} } where l_1, l_2, ..., l_k are some fixed time lags. It will be enough for me to compute

Dear All, Apologies if my questions are too basic for this list. I am given a set of data corresponding to list of detection times (real, non-integer numbers in general) for some events, let us say nuclear decays to fix the ideas. It is a small dataset, corresponding to about 400 nuclear decay times. I would like to test the hypothesis that these decay times are Poissonian-distributed. What is

Dear R-help-archive.. I am trying to figure out how to make arima prediction when I have a process involving multivariate time series input, and one output time series (output is to be predicted) .. (thus strictly speaking its an ARMAX process). I know that the arima function of R was not designed to handle multivariate analysis (there is dse but it doesnt handle arma multivariate analysis, only

Dear useRs, recently, there were several discussions on R-help about how to conveniently fit dynamic linear models and time series regressions. The package dynlm tries to address this problem by 1. providing some more functions like lags L() and differences d() and season() in the formula specification of a model and 2. preserving the time series attributes of the data. A first version of the

Dear useRs, recently, there were several discussions on R-help about how to conveniently fit dynamic linear models and time series regressions. The package dynlm tries to address this problem by 1. providing some more functions like lags L() and differences d() and season() in the formula specification of a model and 2. preserving the time series attributes of the data. A first version of the

Hi Daniel, On Fri, Dec 30, 2016 at 10:55 PM, Daniel Berlin <dberlin at dberlin.org> wrote: >> Right, but we are talking about "when, in the intermediate state, can i >> transform an undef to a different value". >> >> Remember you can only go down the lattice. So you can't make undef >> constant, and then discover it's wrong, and go back up :)

At Mon, 19 Nov 2001 08:36:38, you wrote: > I get the labels I expect: if this is quarterly data the lags are labelled > in years. That is what `frequency = 4' is intended to mean: 4 > observations per unit of time. some further thoughts convinces me that this is a mis-feature. if you ask any person what is the lag i autocorrelation, the answer would be corr(y_t, y_{t-i}). so you

I want to set up a model with a formula and then run dynlm(formula) because I ultimately want to loop over a set of formulas (see end of post) R> form <- gas~price R> dynlm(form) Time series regression with "ts" data: Start = 1959(1), End = 1990(4) <snip> Works OK without a Lag term R> dynlm(gas ~ L(gas,1)) Time series regression with "ts" data: Start =

Hello list, I am trying to test a model but for the beginning I want to do this by using simulated dataset. The model is Y_t = X_t %*% beta + e Where Y : (Nx1); X: (Nxp); beta: (0.6,0.3,0.1); e-uncorrelated normally distributed variates for each t. and later I want to use to use this dataset in a BUGS model to estimate the betas. Thank you for you consideration. [[alternative

Hello, I would like to write a program that compute the principal components of a set of data and then 1. Run the dependent variable against the principal components (lagged value) 2. Do prediction , following Stock and Watson (1999) "Forecasting Inflation". All data are time series. Now I can run the program using single factor (first principal component), but I

Each of the data sets contains monthly observations on price indices for 7 countries. I use the fitted values from reg1 in the reg2 model. The interior loop executes without error as long as I explicitly specify the data set, i.e. data=dat70. However the code fails to execute if I specify the model in the form of the commented line, i. e reg1 <-dynlm(form1,data=Dnames[j]) I get the following

For 2 or 3 stage least squares, what do you R folks do? Follow-up question. My student wants to estimate this. 2 variables are governed by a system of difference equations. His theory is like so. Y_t and X_t are state variables, we want estimates for a, g, b, and h. X_(t+1) = 1 + a X_t + (a/K)* (X_t)^2 - g Y_t X_t Y_(t+1) = b Y_t + h* X_t * Y_t K is perhaps something to estimate, but it

The documentation for nls says the following about the starting values: start: a named list or named numeric vector of starting estimates. Since R 2.4.0, when 'start' is missing, a very cheap guess for 'start' is tried (if 'algorithm != "plinear"'). It may be a good idea to document that when algorithm = "port", if start is a named

I have a question and it's only relation to R is that I probably need R after I understand what to do. Both models are delta y_t = Beta + epslion and suppose I have a null hypothesis and alternative hypothesis H_0 : delta y_t = zero + epsilon epsilon is normal ( 0, sigmazero^2 ) H_1 delta y_t = beta + epsilon epsilon is normal ( sigmabeta^2 )

On Fri, Dec 30, 2016 at 11:55 PM, Sanjoy Das <sanjoy at playingwithpointers.com > wrote: > Hi Daniel, > > On Fri, Dec 30, 2016 at 10:55 PM, Daniel Berlin <dberlin at dberlin.org> > wrote: > >> Right, but we are talking about "when, in the intermediate state, can i > >> transform an undef to a different value". > >> > >>

Dear list, I posted essentially this same question to the r-sig-geo mailing list last week with no response :( Unfortunately I am no closer to reaching a solution, so I now post it here (with some clarifications) in the hope that someone following this list might have an answer for me: Has anyone had much experience with spatial interaction (or gravity) models, specifically in the form of