Discrete Choice
How is discrete choice different from classical demand forecasting?
Classical forecasting predicts quantity; discrete choice predicts the probability of selecting one option from a finite set, which is better suited to supplier selection, mode choice, and substitution decisions.
What is the basic statistical form of a discrete choice model?
The model typically specifies latent utility for each alternative as a function of alternative attributes and decision-maker characteristics, then maps utilities into choice probabilities; common forms include logit-family models.
Why is discrete choice useful in operations management?
It turns qualitative operational decisions into measurable probabilities, allowing planners to evaluate how lead time, cost, quality, and service-level changes shift choice behavior across procurement and inventory policies.
Discrete choice is a modeling framework for decisions among a finite set of mutually exclusive alternatives, where an actor selects one option from a choice set rather than a continuous quantity. In supply chain, it estimates which supplier, SKU, replenishment option, transport mode, or service level a firm chooses based on observable attributes and latent utility.
In a manufacturing shop floor or warehouse, discrete choice models apply when decisions are inherently categorical: selecting one carrier, one vendor, one lot, one machine setup, or one reorder policy from several alternatives. Each alternative's attractiveness is expressed as a function of attributes such as lead time, price, defect rate, fill rate, MOQ, distance, or handling cost, and the model estimates the probability that a planner or system chooses each option. In procurement, this supports supplier selection by quantifying trade-offs among cost, delivery reliability, quality, and risk rather than treating the decision as simple lowest-price. In inventory and material tracking, discrete choice represents whether to expedite, backorder, substitute, split-ship, or hold production when a shortage occurs, allowing planners to evaluate policy impacts under different states. In operations research, discrete choice models are embedded in optimization so supply and demand decisions are solved jointly and remain linear in some formulations.
Stockouts from overestimated substitution: Assuming planners switch materials more often than they do understates primary SKU demand, causing repeated shortages on the exact item the line needs.
Receiving bottlenecks from poor choice-set definition: Omitting realistic options like partial receiving or carrier splits misrepresents receiving decisions and distorts dock, labor, and storage planning.
Supplier-selection errors from ignoring non-price attributes: Optimizing only unit cost while excluding defect rate, lateness, or variability leads to cheap-looking suppliers that cause line stoppages, expediting, and excess safety stock.