The amount of shrinkage for parameters varies by parameter selection, but the presence of collinearity is required for the weights to shrink. You can demonstrate this to yourself by using truly random (IID explanatory variables generated by a random generator) versus features that are highly dependent on one another (for example, waist line and weight).
Here are two examples of extreme regularization values and their effect on model weights and shrinkage:
val regularizationParam = .00001(Model Weights:,[-0.0373404807799996, 0.25499013376755847, 0.0049174051853082094, 0.0046110262713086455, 0.027391063252456684, 0.6401656691002464, 0.1911635644638509, 0.4085780172461439 ]) val regularizationParam = 50(Model Weights:,[-0.012912409941749588, ...
