vault / data / optimization.md

Mixed-integer optimization

Data & exploration 72 words 1 outgoing 0 backlinks

Assign discrete resources - which truck, which delivery slot - under constraints, with a cost to minimize. I run the ops side of one in production: CVXPY to prototype the model, Kedro to keep the pipeline reproducible. Gotchas: the model is the easy half, and bad input data surfaces as infeasible at 6am; always keep a fallback plan for solver timeouts, and log the optimality gap, not just the objective.

Reference: CVXPY.

LINKS IN THIS NOTE

[[ Kubeflow Pipelines ]] Kubeflow Pipelines turns train-test-deploy into a DAG of containerized components on Kubernetes, each with explicit inputs and outputs. It earns its weight when pipelines are long, rerunnable and shared; a two-step nightly job is happier as a script. Gotchas: a...