missoNet

Efficient procedures for fitting conditional graphical lasso models that link a set of predictor variables to a set of response variables (or tasks), even when the response data may contain missing values. 'missoNet' simultaneously estimates the predictor coefficients for all tasks by leveraging information from one another, in order to provide more accurate predictions in comparison to modeling them individually. Additionally, 'missoNet' estimates the response network structure influenced by conditioning predictor variables in a L1-regularized conditional Gaussian graphical model. Unlike most penalized multi-task regression methods (e.g., MRCE), 'missoNet' is capable of obtaining estimates even when the response data is corrupted by missing values. The method automatically enjoys the theoretical and computational benefits of convexity, and returns solutions that are comparable to the estimates obtained without any missing values. The package also includes auxiliary functions for data simulation, adaptive model initialization, hyper-parameter tuning, goodness-of-fit evaluation, and visualization of estimates, as well as predictions in new data.