Location Optimizer
The Location Optimizer step chooses K sites from a table of candidate locations that minimize the total or weighted distance to a table of demand points — a p-median facility-location model solved as a mixed-integer program on PlaidCloud’s HiGHS solver.
Location Optimizer runs in a workflow job pod, the same job image Solver uses, and writes results back to project tables.
Inputs
Section titled “Inputs”Candidate Sites
Section titled “Candidate Sites”The table of possible site locations, mapped with the same source/target column mapping, filters, and aggregation every table-mapped input uses.
Demand Points
Section titled “Demand Points”The table of points whose distance to the nearest selected site is minimized, mapped the same way.
Distance Table (Optional)
Section titled “Distance Table (Optional)”A precomputed table of candidate-to-demand distances. Leave unset to have the step compute distances itself.
The number of sites to select from Candidate Sites.
Solver Options
Section titled “Solver Options”| Option | Description |
|---|---|
| Time Limit | Optional runtime limit in seconds. |
| MIP Gap | Optional relative MIP gap. |
| Enable Presolve | Let the solver simplify the model before solving. Enabled by default. |
| Use Custom Resources | Set CPU cores and memory for the job pod. |
A Time Limit or MIP Gap stop returns the best solution found within that limit rather than the true optimum — the same trade-off Solver makes on a non-optimal status. Leave both unset to solve to full optimality.
Outputs
Section titled “Outputs”Location Optimizer writes three output tables, which must all be named distinctly.
Selected Sites Output Table
Section titled “Selected Sites Output Table”One row per chosen site.
Assignment Output Table
Section titled “Assignment Output Table”One row per demand point, naming the selected site it was assigned to.
Summary Output Table
Section titled “Summary Output Table”One row with solver status, objective value, and run details.
Limits and Caveats
Section titled “Limits and Caveats”This step is a p-median model: it always minimizes total (optionally weighted) distance from demand points to their nearest selected site. It does not express a gravity-style score, a custom scoring formula, or any objective where a higher score is better rather than a lower distance. A site-selection problem that needs one of those belongs in a hand-built Solver model instead.
Related
Section titled “Related”- Solver — general linear and mixed-integer optimization, for a site-selection objective this step’s fixed p-median model doesn’t cover.
- Alteryx Conversion Matrix — the Alteryx Location Optimizer tool converts to this step only for its linear distance-minimization case; a gravity kernel or custom scoring macro still refuses.