Application of Mathematical Optimization Procedures to Intervention Effects in Structural Equation Models

TL;DR

This paper introduces a convex quadratic programming approach to optimize interventions in structural equation models, aiming to minimize output variances and adjust expectations effectively.

Contribution

It develops a novel optimization framework for interventions in SEMs, incorporating boundary conditions and target adjustments, enhancing control over model outputs.

Findings

The intervention problem is formulated as a convex quadratic programming task.

The method allows practical boundary conditions for interventions.

The approach effectively reduces variances and adjusts expectations in SEMs.

Abstract

For a given statistical model, it often happens that it is necessary to intervene the model to reduce the variances of the output variables. In structural equation models, this can be done by changing the values of the path coefficients by intervention. First, we explain that the expectations and variance matrix can be decomposed into several parts in terms of the total effects. Then, we show that an algorithm to obtain intervention method which minimizes the weighted sum of the variances can be formulated as a convex quadratic programming. This formulation allows us to impose boundary conditions for the intervention, so that we can find the practical solutions. We also treat a problem to adjust the expectations on targets.