Methods and apparatus for scheduling timeslots using machine learning processes

Patent metadata

Field	Value
Publication number	US-2023419269-A1
Application number	US-202318455944-A
Country	US
Kind code	A1
Filing date	Aug 25, 2023
Priority date	Jan 29, 2021
Publication date	Dec 28, 2023
Grant date	—

Abstract

This application relates to automatically scheduling timeslots, such as timeslots for scheduling item pickups and deliveries. The embodiments may employ machine learning process to determine, for each of a plurality of timeslots, a timeslot capacity. The embodiments may further determine a time as to when each of the plurality of timeslots become available for selection, such as on a webpage. In some examples, the time determined to make available for selection each of the plurality of timeslots is based on a predetermined amount of time before the timeslot. Further, the embodiments may provide for display each of the plurality of timeslots at the determined time. The embodiments may further include receiving a selection of one of the plurality of timeslots, and scheduling a pickup or delivery based on the selected timeslot. The machine learning processes may be trained with features generated from historical timeslot data and workforce availability.

First claim

1 . A system comprising:
a computing device comprising a memory and at least one processor, wherein the at least one processor is configured to:
receive a request for capacity information about a plurality of timeslots in a first future period, wherein the first future period is associated with an online order placed by a user;
obtain order data associated with the plurality of timeslots in a first previous period;
obtain workforce availability data for a store associated with the online order for the plurality of timeslots in the first future period;
generate first features based on the order data;
generate second features based on the workforce availability data;
determine, for each timeslot in a major portion of the plurality of timeslots in the first future period, a timeslot capacity based on applying a machine learning model to the first features and the second features;
determine, for each timeslot in the remaining portion of the plurality of timeslots in the first future period, a first increased timeslot capacity higher than the timeslot capacity based on a reinforcement learning model.

2 . The system of claim 1 , wherein the machine learning model is a time series model.

3 . The system of claim 1 , wherein the machine learning model is trained based on:
obtaining additional order data for a second previous period; obtaining additional workforce availability data for the store for the second previous period; generating feature vectors based on the additional order data and the additional workforce availability data; and training the machine learning model with the feature vectors.

4 . The system of claim 1 , wherein the at least one processor is further configured to:
determine a timeslot availability time for each of the plurality of timeslots based on a predetermined amount of time before each timeslot, wherein the timeslot data comprises the timeslot availability times; and provide for display the timeslot availability time for each of the plurality of timeslots.

5 . The system of claim 1 , wherein the at least one processor is further configured to determine a timeslot availability time for each of the plurality of timeslots based on the workforce availability data, wherein the timeslot data comprises the timeslot availability times.

6 . The system of claim 1 , wherein the at least one processor is further configured to:
generate timeslot data comprising: (a) the timeslot capacities for the major portion of the plurality of timeslots in the first future period, and (b) the first increased timeslot capacities for the remaining portion of the plurality of timeslots in the first future period; transmit the timeslot data to be displayed to the user for scheduling a pickup or delivery of the online order during one of the plurality of timeslots in the first future period; identify whether the first increased timeslot capacity is met based on order data in the first future period; and generate an updated timeslot capacity.

7 . The system of claim 6 , wherein the updated timeslot capacity is generated based on whether the first increased timeslot capacity is met.

8 . The system of claim 6 , wherein the at least one processor is further configured to:
re-train the machine learning model with feature vectors generated based on the order data in the first future period, using the updated timeslot capacity as a ground truth; generate a predicted timeslot capacity based on the re-trained machine learning model; determine a second increased timeslot capacity higher than the predicted timeslot capacity based on the reinforcement learning model; generate updated timeslot data comprising: (c) the predicted timeslot capacities for a major portion of a second future period, and (d) the second increased timeslot capacities for the remaining portion of the second future period; and transmit the updated timeslot data to be displayed to users for scheduling pickup or delivery of online orders in the second future period.

9 . A method, implemented by a computing device comprising a memory and at least one processor, comprising:
receiving a request for capacity information about a plurality of timeslots in a first future period, wherein the first future period is associated with an online order placed by a user; obtaining order data associated with the plurality of timeslots in a first previous period; obtaining workforce availability data for a store associated with the online order for the plurality of timeslots in the first future period; generating first features based on the order data; generating second features based on the workforce availability data; determining, for each timeslot in a major portion of the plurality of timeslots in the future period, a timeslot capacity based on applying a machine learning model to the first features and the second features; determining, for each timeslot in the remaining portion of the plurality of timeslots in the first future period, a first increased timeslot capacity higher than the timeslot capacity based on a reinforcement learning model.

10 . The method of claim 9 , further comprising:
determining a timeslot availability time for each of the plurality of timeslots based on a predetermined amount of time before each timeslot, wherein the timeslot data comprises the timeslot availability times; and providing for display the timeslot availability time for each of the plurality of timeslots.

11 . The method of claim 9 , further comprising determining a timeslot availability time for each of the plurality of timeslots based on the workforce availability data, wherein the timeslot data comprises the timeslot availability times.

12 . The method of claim 9 , further comprising:
generating timeslot data comprising: (a) the timeslot capacities for the major portion of the plurality of timeslots in the first future period, and (b) the first increased timeslot capacities for the remaining portion of the plurality of timeslots in the first future period; transmitting the timeslot data to be displayed to the user for scheduling a pickup or delivery of the online order during one of the plurality of timeslots in the first future period; identifying whether the first increased timeslot capacity is met based on order data in the first future period; and generating an updated timeslot capacity.

13 . The method of claim 12 , wherein the updated timeslot capacity is generated based on whether the first increased timeslot capacity is met.

14 . The method of claim 12 , further comprising:
re-training the machine learning model with feature vectors generated based on the order data in the first future period, using the updated timeslot capacity as a ground truth; generating a predicted timeslot capacity based on the re-trained machine learning model; determining a second increased timeslot capacity higher than the predicted timeslot capacity based on the reinforcement learning model; generating updated timeslot data comprising: (c) the predicted timeslot capacities for a major portion of a second future period, and (d) the second increased timeslot capacities for the remaining portion of the second future period; and transmitting the updated timeslot data to be displayed to users for scheduling pickup or delivery of online orders in the second future period.

15 . A non-transitory computer readable medium having instructions stored thereon, wherein the instructions, when executed by at least one processor, cause a device to perform operations comprising:
receiving a request for capacity information about a plurality of timeslot

Assignees

• Walmart Apollo Llc

Inventors

• Vasagiri Ajaykartheek
• Hussain Faraz
• Manglani Kamal
• Sureshkumar Ashwin
• Krishnaswamy Sriram
• Tripathy Chittaranjan

Classifications

• G06Q10/1093 Primary

  Calendar-based scheduling for persons or groups · CPC title

• G06N20/00

  Machine learning · CPC title

• G06Q50/30

  Physics · mapped topic

• G06Q10/0835

  Relationships between shipper or supplier and carriers · CPC title

• G06N20/20

  Ensemble learning · CPC title