We consider the penalized security-constrained optimal power flow (SCOPF) problem in a linearized form where thermal line limits are enforced as soft constraints to reflect operational flexibility. Unlike traditional screening, we characterize constraint redundancy as a load-dependent property of penalized optima. We propose a constraint-reduction method for the penalized SCOPF based on a nested hierarchy of non-redundant constraint subsets, each hierarchical level defining an equivalent problem involving fewer constraints. We develop an extraction algorithm to construct this hierarchy from the unpenalized problem and relate load profiles to a specific level using our characterization. We introduce a cumulative encoding coupled with a low-dimensional supervised learning approach to predict the hierarchy level corresponding to a load profile for the penalized SCOPF. Exactness is guaranteed through an optional correction step. Using LightGBM predictors on the IEEE 39-bus system, we achieve a 94.28% hierarchy level prediction accuracy,