Figure 1
From: Differential expression analysis with global network adjustment
Network depiction of regression estimation. The regression coefficients are shown as rays from the origin. The absolute value of the coefficient is the distance from the origin and the sign is indicated by blue (+) or red (-). There are 200 rays indicating 200 estimated coefficients, and the true values of the coefficients (+1,-1) appear as rings. The Ordinary Least Squares (OLS) estimates are highly variable because of the small sample size and the correlation between the positive and negative predictor groups. The elastic net has less variable estimates than OLS, but the OSRR coefficients demonstrate the the strong smoothing property of that equalizes the influence of predictors.