Fig. 1From: Lasso regularization for left-censored Gaussian outcome and high-dimensional predictorsObserved and predicted simulated censored data. Regression lines are obtained using: the true model that generated the data, the gold standard, maximum likelihood estimation (MLE), non-parametric Buckley-James (BJ), least absolute deviations (LAD) and censored LAD regressions, simple imputation by the limit of detection (LOD) and by LOD /2Back to article page