Table-turn prediction from POS event timing

What is claimed is:

1. 1.
   Independent · Method

   A computer-implemented method of predicting when a table in a restaurant will become available, the method comprising: receiving, from a point-of-sale system, a stream of timestamped events for an open check associated with the table, the events including at least an order event and a payment event; determining, from the timestamped events, a current course stage of the open check; estimating, by a prediction model, a remaining occupancy time for the table based on the current course stage, an elapsed time since the order event, and a party size of the open check; and transmitting, to a host interface, a predicted availability time for the table derived from the remaining occupancy time.

2. 2.

   The method of claim 1, wherein the timestamped events further include a course-fire event sent from the point-of-sale system to a kitchen display system, and wherein determining the current course stage comprises identifying the most recent course-fire event.

3. 3.

   The method of claim 1, wherein the prediction model is trained on historical checks of the restaurant, each historical check labeled with the time between its order event and the closing of the check.

4. 4.

   The method of claim 3, further comprising retraining the prediction model at a fixed interval using checks closed since the previous training.

5. 5.

   The method of claim 1, further comprising updating the predicted availability time each time a new timestamped event is received for the open check.

6. 6.

   The method of claim 1, wherein estimating the remaining occupancy time further comprises applying a correction for the day of the week and the hour of the day of the open check.

7. 7.

   The method of claim 1, further comprising, in response to receiving the payment event, reducing the remaining occupancy time to a clearing interval measured for a section of the restaurant that contains the table.

8. 8.

   The method of claim 1, further comprising ranking waiting parties against tables by matching a party size of each waiting party to the predicted availability times, and displaying the ranking on the host interface.

9. 9.
   Independent · System

   A system for predicting table availability, the system comprising: one or more processors; and memory storing instructions that, when executed by the one or more processors, cause the system to: receive, from a point-of-sale terminal, timestamped events for an open check associated with a table; determine a current course stage of the open check from the timestamped events; estimate, with a prediction model, a remaining occupancy time for the table from the current course stage, an elapsed time and a party size; and display, on a host device, a predicted availability time for the table.

10. 10.

    The system of claim 9, wherein the instructions further cause the system to receive course-fire events from a kitchen display system and to use the most recent course-fire event to determine the current course stage.

11. 11.

    The system of claim 9, wherein the prediction model is a gradient-boosted regression model trained on historical checks of the restaurant.

12. 12.

    The system of claim 11, wherein the historical checks are weighted so that more recent checks contribute more to training.

13. 13.

    The system of claim 9, wherein the host device displays each table with its predicted availability time and a confidence interval around the predicted availability time.

14. 14.

    The system of claim 9, wherein the instructions further cause the system to flag the table when the elapsed time exceeds the estimated occupancy by more than a set margin.

15. 15.

    The system of claim 9, wherein the instructions further cause the system to quote a wait time to a waiting party from the earliest predicted availability time among tables that seat the waiting party.

16. 16.
    Independent · Non-transitory computer-readable medium

    A non-transitory computer-readable medium storing instructions that, when executed by a processor, cause the processor to perform operations comprising: receiving timestamped point-of-sale events for an open check associated with a table; determining a current course stage of the open check from the events; estimating a remaining occupancy time for the table from the current course stage, an elapsed time and a party size; and outputting a predicted availability time for the table to a host interface.

17. 17.

    The medium of claim 16, wherein the operations further comprise updating the predicted availability time on each new event for the open check.

18. 18.

    The medium of claim 16, wherein the events include a payment event, and wherein the operations further comprise reducing the remaining occupancy time to a clearing interval upon the payment event.

19. 19.

    The medium of claim 16, wherein the operations further comprise quoting a wait time to a waiting party based on the predicted availability time.

20. 20.

    The medium of claim 19, wherein the wait time is sent to the waiting party as a text message.