automatic general to specific regression with sign constraints on coefficients
[rd,rf]=automatic_signed(namey,listx,signx,arg1,...,argn)
* namey = a time series, a real (nx1) vector or a string equal to the name of a time series or a (nx1) real vector between quotes* results = an automatic tlist results
* listx = a matrix of exogenous variables, under as a (T x k) real matrix or a (T x 1) string vector of names
* signx = a (T x 1) vector of -1, 1 or %nan values, each value indicating the expected sign of the coefficients, -1 for negative, +1 positive and %nan for indifferent.
* arg1,...,argn = any option to automatic (see function automatic for details)
* rd = a results tlist with:
- rd('meth') = 'signed automatic'
- rd('starting automatic') = original automatic results tlist
- rd('ending automatic') = final automatic results tlist that is with all coefficient fo the good sign
- rd('removed var') = a string vector, the variables that have been removed
* rf = the results tlist of the final model
load(GROCERDIR+'\data\conjfra.dat') bounds('1991q3','2008q2','2009q3','2011q2') tests='test=predfailin(0.9),doornhans,arlm(4),hetero_sq' signx=ones(size(listx,2),1) [r1,rf1]=automatic_signed('growthr(fra_gdp)',listx,signx,'comp=const','strategy=liberal',tests) // Example taken from function automatic_signed_d. Sign of all free coefficients are imposed positive. quantile(x,[0.025 ; 0.5 ; 0.975]) | ![]() | ![]() |