zhan-gao / classo

A package implements Classifier-Lasso

Welcome to my Github profile page 👋

I am a Ph.D. student in the Department of Economics, University of Southern California. My research interests are mainly in econometrics.

🔭 I’m currently working on high-dimensional predictive regression models, random coefficient models and panel data models with grouped heterogeneity. Feel free to check my research profile and CV out.

🌱 I'm a R user and maintaining classo and LasForecast.

💬 zhangao [at] usc [dot] edu

classo's Issues

A reminder to myself

A few things to improve:

1. SPJ_PLS cannot handle shared trends for each i. They singular after demeaning.
2. write the help document for data.normalization.

Encounter errors using functions 'PLS.mosek' and "PLS.nlopt"

I generate data by DGP 1 described in Su, Shi and Phillips (2016) and it works well using "PLS.cvxr". However, when I use "PLS.mosek" with the code
pls_out <- PLS.mosek(n, tt, y, x, K = 3, lambda = lambda)
, it raise error:
Error in a.out[k, ] <- mosek.out$alpha :
number of items to replace is not a multiple of replacement length

When I use "PLS.nlopt" with
pls_out <- PLS.nlopt(n, tt, y, x, K = 3, lambda = lambda)
, it raise error:
Error in eval_f(x0, ...) : object 'p' not found

I have checked that the package "Rmosel" and "nloptr" are rightly installed.

mosek_version()
[1] "MOSEK 10.0.46"

remote installation failed on windows with R 3.5

solved by cloning and install from local place.