US 12,271,941 B2
Machine learning methods for commercial lease benchmarking and devices thereof
Pradnya Nimkar, Newark, CA (US); Utkarsh Porwal, Santa Clara, CA (US); and Gurdit Chahal, Elk Grove, CA (US)
Assigned to JONES LANG LASALLE IP, INC., Chicago, IL (US)
Filed by JONES LANG LASALLE IP, INC., Chicago, IL (US)
Filed on Jan. 12, 2023, as Appl. No. 18/096,472.
Prior Publication US 2024/0242266 A1, Jul. 18, 2024
Int. Cl. G06Q 30/00 (2023.01); G06Q 30/0645 (2023.01); G06Q 50/163 (2024.01)
CPC G06Q 30/0645 (2013.01) [G06Q 50/163 (2013.01)] 12 Claims
OG exemplary drawing
 
1. A method for improved machine learning modeling to facilitate automated commercial lease benchmarking, the method implemented by a property analysis server device and comprising:
training, by one or more processors of the property analysis server device, machine learning models (MLMs) on a feature dataset comprising property data for properties and metric data, wherein the property data comprises at least an address and an actual lease value for each of the properties, each of the MLMs comprises a different MLM type, and the property data is retrieved via a wide area network from one or more property data servers;
cross-validating a selected one of the MLMs comprising iteratively retraining the selected one of the MLMs using subsets of the feature dataset, generating predicted lease values for the properties, and comparing the actual lease values to the predicted lease values to determine that the selected one of the MLMs exceeds an accuracy threshold;
storing the property data in a lease benchmarking database in a memory of the property analysis server device, wherein the addresses are replaced in the stored property data with corresponding first geographic coordinates and geohash values and the properties are associated in the lease benchmarking database with the predicted lease values;
sending via the wide area network an address included in a lease pricing request to a maps platform server based on a provided application programming interface (API) to obtain second geographic coordinates for the address, wherein the lease pricing request is received from a user device via an enterprise network coupled to the wide area network;
generating a haversine distance to a subset of the properties associated with a geohash value in the lease benchmarking database to identify one of the subset of the properties, wherein the geohash value is identified based on the second geographic coordinates; and
returning to the user device via the enterprise network and the wide area network one of the predicted lease values in response to the lease pricing request, wherein the one of the predicted lease values is identified in the lease benchmarking database based on a determined geographic proximity of the address to the one of the subset of the properties.