binom | Quantlet to compute binomial coefficient |
cafpe |
Quantlet to conduct lag selection for the conditional
mean function in nonlinear autoregressive
models. It also allows for prior data transformations.
It uses local linear estimation for the
estimation of a corrected Asymptotic Final Prediction Error
(CAFPE).
This quantlet does not allow to change advanced
parameter settings nor to select lags for the conditional
volatility function. For doing this, use the quantlet cafpefull.
|
cafpedefault | Quantlet to define advanced parameters for conducting lag selection for nonlinear autoregressive models using the quantlet cafpefull |
cafpefull | Quantlet to conduct lag selection for conditional mean or conditional volatility function of nonlinear autoregressive models. It also allows for prior data transformations. It can be based on either the local linear estimation of the Asymptotic Final Prediction (AFPE) or a corrected version (CAFPE). However, only for CAFPE the used plug-in bandwidth is consistent. |
cafpeload |
Quantlet to load all quantlets including dll/so files
which are necessary to run
nonparametric lag selection and nonparametric nonlinear
autoregression.
|
dencp |
Quantlet for multivariate density estimation using
kernel estimation using C++ routines via a DLL.
|
fgrfsv |
Quantlet to compute local linear estimator of conditional
mean function
|
fpefsv |
Quantlet to compute (C)AFPE given estimates for B and C and the
asymptotically optimal bandwidth. If a scalar bandwidth is given
it computes AFPE, CAFPE using this bandwidth. If a vector bandwidth
is given, it only computes Ahat and the residuals.
|
fpenps |
Quantlet to conduct lag selection for nonlinear autoregressive
models. It can be based on either the
local linear estimation of the Asymptotic Final Prediction
(AFPE) or a corrected version (CAFPE)
|
fpenpsl | Quantlet to compute lag selection criteria for nonlinear autoregressive models for a given vector of lags. It allows to compute two criteria based on local linear estimation of the Asymptotic Final Prediction Error: AFPE and CAFPE. If a scalar bandwidth is given, it is used as hA in the computation of AFPE and CAFPE. If a vector bandwidth is given, only the residuals are computed and zeros returned for the criteria. |
fvllc |
Quantlet for multivariate local linear or partial local quadratic
estimation using C++ - routines via dlls.
It can estimate conditional means, conditional volatilities,
first derivatives, second direct derivatives,
conditional densities with full or
leave-one-out and density estimation with all possible
data or only lagged data.
Only the gaussian kernel can be used.
|
hoptest |
Quantlet for computing the scalar plug-in bandwidth
for nonparametric estimation of nonlinear autoregression
models of order p. The unknown quantities in the
asymptotic optimal bandwidth are nonparametrically
estimated using C++ - routines via dlls.
Only the gaussian kernel can be used.
|
hsilv | Quantlet to compute Silverman's rule-of-thumb bandwidth for density estimation using either the gaussian or uniform kernel |
lagdir1 | Quantlet for conducting directed search as in Tjostheim and Auestad (1994) for lag selection. It collects all lags which have not yet been selected into a vector lagno |
lagfull | Quantlet to generate matrix of all possible combinations of lags in order to conduct a full search for lag selection. |
makegrid | Quantlet for generating a two-dimensional grid |
plotloclin | Quantlet to compute for a given lag vector on given grid range a 1- or 2-dimensional plot of the regression function of a nonlinear autoregressive process; if more than 2 lags are used, then only two lags are allowed to vary, the others have to be fixed at values that are given by the user; the procedure uses a plug-in bandwidth; for this bandwidth the vector of residuals and the matrix of regressors are returned on which the bandwidth estimation was conducted (for the default values of the advanced parameters see the quantlet cafpedefault); plots also standardized residuals |
plotoneline | Quantlet to produce a graph of a matrix |
wei |
Quantlet for weighting the observations.
It returns either a 1 or 0 for each observation
according to whether it should be kept or ignored.
This is decided on whether the density of the vector
of lags is larger than a certain threshold. The threshold is computed
such that per*#obs observations are thrown away
which are those with the lowest density.
If a density matrix is handed over, the operation is
done on column i.
|
xorigex | Quantlet to construct matrix of lagged variables |
xorigst | Quantlet to cut off starting values from matrices of dependent and independent variables for time series analysis. This is needed if one always wants to cut off the same number of observations when analysing AR(p) models with different orders. |