Ycriteria

Ycriteria is a term used to describe a family of evaluation criteria that are parameterized by a real-valued variable y. In this sense, for each y in a set Y there is a criterion C_y: X → R that assigns a score to each candidate x in a domain X. The collection of these scoring rules is referred to as the ycriteria. The parameter y can encode preferences, risk tolerance, or thresholds, so changing y yields different rankings or selections from the same domain.

Common forms include linear combinations, such as C_y(x) = w1(x) + y·w2(x), where y modulates the relative weight of two components. Another form is loss-plus-penalty, C_y(x) = Loss(x) + y·Penalty(x), with y controlling the trade-off between fidelity and complexity. In thresholding scenarios, y may specify a probability or percentile whose associated criterion reorders options. Analysts may study how C_y(x) varies with y to understand the stability and sensitivity of decisions across different operating conditions.

Properties of interest include monotonicity or continuity in y, which affect the stability of selections as y changes. If the family is convex in x for every y, standard optimization techniques can be applied; if C_y is convex in y for fixed x, sensitivity analysis becomes more straightforward.

Applications of ycriteria appear in parameterized model evaluation, decision analysis, and multi-criteria optimization, where different stakeholders or operating conditions require different evaluation emphasis. They provide a unified framework to compare alternatives under a continuum of criteria rather than a single metric. The term is not universally standardized and is often described as describing a parametric evaluation framework within context-specific literature.