Approximation Algorithm for Generalized Budgeted Assignment Problems and Applications in Transportation Systems

TL;DR

This paper introduces a novel randomized approximation algorithm for a complex capacitated assignment problem with budget constraints, motivated by transit planning, and demonstrates its effectiveness through real-world data experiments.

Contribution

It presents the first constant-ratio randomized rounding algorithm for the generalized budgeted assignment problem and applies it successfully to transportation system planning.

Findings

The algorithm achieves a constant approximation ratio.

Numerical experiments show significant improvements over previous methods.

Application to transit planning demonstrates practical effectiveness.

Abstract

Motivated by a transit line planning problem in transportation systems, we investigate the following capacitated assignment problem under a budget constraint. Our model involves $L$ bins and $P$ items. Each bin $l$ has a utilization cost $c_l$ and an $n_l$-dimensional capacity vector. Each item $p$ has an $n_l$-dimensional binary weight vector $r_{lp}$, where the $1$s in $r_{lp}$ (if any) appear in consecutive positions, and its assignment to bin $l$ yields a reward $v_{lp}$. The objective is to maximize total rewards through an assignment that satisfies three constraints: (i) the total weights of assigned items do not violate any bin's capacity; (ii) each item is assigned to at most one open bin; and (iii) the overall utilization costs remain within a total budget $B$. We propose the first randomized rounding algorithm with a constant approximation ratio for this problem. We then apply our framework to the motivating transit line planning problem, presenting corresponding models and conducting numerical experiments using real-world data. Our results demonstrate significant improvements over previous approaches in addressing this critical transportation challenge.