Keywords - Function groups - @ A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

Library: cafpe

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.

Keywords - Function groups - @ A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

(C) MD*TECH Method and Data Technologies, 27.4.2000