期刊名称：CREED Working Papers / Center for Research in Experimental Economics and Political Decision-Making
出版年度：2014
出版社：Amsterdam
摘要：We develop an algorithm that incorporates network information into regressionsettings. It simultaneously estimates the covariate coefficients and the signsof the network connections (i.e. whether the connections are of an activating or ofa repressing type). For the coefficient estimation steps an additional penalty is seton top of the lasso penalty, similarly to Li and Li (2008). We develop a fast implementationfor the new method based on coordinate descent. Furthermore, we showhow the new methods can be applied to time-to-event data. The new method yieldsgood results in simulation studies concerning sensitivity and specificity of non-zerocovariate coefficients, estimation of network connection signs, and prediction performance.We also apply the new method to two microarray time-to-event datasets from patients with ovarian cancer and diffuse large B-cell lymphoma. The newmethod performs very well in both cases. The main application of this new methodis of biomedical nature, but it may also be useful in other fields where network datais available
