Automatically determining market rental rates for properties

• US 10,460,406 B1
• Filed: 03/09/2011
• Issued: 10/29/2019
• Est. Priority Date: 03/09/2011
• Status: Active Grant

• Current Assignee: MFTB Holdco Incorporated
• Original Assignee: Zillow, Inc. (Zillow Group, Inc.)
• Inventors: Humphries, Stanley B., Xiang, Dong, Bun, Yeng
• Primary Examiner(s): Gilkey, Carrie S
• Application Number:
  US13/044,480
• Time in Patent Office: 3,156 Days
• Field of Search: 705306, 705313
• CPC Class Codes: G06Q 30/0645 Rental transactions; Leasing transactions
• CPC Class Codes: G06Q 50/16 Real estate
• CPC Class Codes: G06Q 50/163 Real estate management

Abstract

A facility for determining a market rental rate for a distinguished home located in a distinguished geographic area is described. The facility receives home attributes for the distinguished home. The facility obtains a market rental rate for the distinguished home by applying to the received home attributes a valuation model for homes in the distinguished geographic area that has been trained using rental listing price and home attribute data from homes recently listed for rental in the distinguished geographic area. The facility reports the obtained market rental rate for the distinguished home.

Claims

1. A method for estimating a market rental rate for a home located in a specific geographic area, comprising:

retrieving data for the specific geographic area, the data comprising a plurality of entries each for a home located in the specific geographic area that was listed for rental after a specific date, each entry indicating (1) a rental listing price for the home, and, (2) values for each of a plurality of attributes including (a) rental attributes, (b) property attributes, and (c) an estimated home value attribute;

training, by a computer, a forest of classification trees to estimate a market rental rate for a home in the specific geographic area using the data retrieved for the specific geographic area;

receiving attributes of a plurality of homes in the specific geographic area;

periodically generating a market rental rate index for the specific geographic area at least in part by, for each home of the plurality of homes in the specific geographic area, applying the trained forest of classification trees to the received attributes of the home to determine a market rental rate for the home;

determining a market rental rate index for the specific geographic area using the market rental rates determined for each home of the plurality of homes in the specific geographic area; and

causing the determined market rental rate index for the specific geographic area to be provided to a user to enable the user to make a determination regarding a home in the specific geographic area.

2. The method of claim 1, wherein determining the market rental rate of the home comprises:

for each classification tree of the trained forest of classification trees,

identifying a leaf node of the classification tree representing attribute ranges containing the home'"'"'s attributes, and

determining a weight for the classification tree based on its score; and

determining the market rental rate for the home based at least in part on an aggregation of the prices of the identified leaf nodes in the weighted classification trees.

3. The method of claim 2, wherein the aggregation determines a mode of the market rental rates.

4. The method of claim 2, wherein the aggregation determines a percentile of the market rental rates.

5. The method of claim 2 wherein the aggregation determines a median of the market rental rates.

6. The method of claim 1, further comprising smoothing out a plurality of market rental rate indices periodically generated for the specific geographic area, wherein the plurality of market rental rate indices include at least a first generated market rental rate index and a second generated market rental rate index, wherein the first generated market rental rate index is different from the second generated market rental rate index.

7. The method of claim 6 wherein smoothing out a plurality of market rental rate indices generated for the specific geographic area comprises:

for each of a plurality of subcategories of homes within the specific geographic area,

determining a raw median value for the subcategory of homes within the specific geographic area,

counting the homes included in the subcategory of homes within the specific geographic area,

determining a weight to apply to the raw median value based on the count of homes included in the subcategory of homes within the specific geographic area.

8. The method of claim 6 wherein the plurality of subcategories of homes within the specific geographic area include price tiers, all homes, single family homes, condos and co-ops, one bedroom homes, two bedroom homes, three bedroom homes, four bedroom homes, and five-plus bedroom homes.

9. A method in a computing system for estimating a market rental rate for a specific home located in a specific geographic area, comprising:

identifying listing information for each of a plurality of homes in the specific geographic area listed for rental after a specific past date, the listing information including (1) values for each of;

(a) at least one property attribute, (b) an estimated home value attribute, (c) at least one rental attribute, and (2) a rental listing price for each home;

training, by a computer, an initialized market rental rate model to estimate a market rental rate for a home in the specific geographic area using the identified listing information;

periodically generating a market rental rate index for the specific geographic area at least in part by, for each home of the plurality of homes in the specific geographic area, applying the trained market rental rate model to attributes of the home to determine a market rental rate for the home;

determining a market rental rate index for the specific geographic area using the market rental rates estimated for each home of the plurality of homes in the specific geographic area; and

causing a determined market rental rate for a specific home to be provided to a user to enable the user to make a determination regarding the specific home.

10. The method of claim 9 wherein the market rental rate model is trained using the identified listing information for a first fraction of homes in the specific geographic area listed for rental after the specific past date, the method further comprising:

selecting a second fraction of homes in the specific geographic area listed for rental after the specific past date that is distinct from the first fraction; and

using identified listing information for the second faction of homes to validate the trained market rental rate model by determining the extent to which the trained market rental rate model'"'"'s market rental rates for the homes of the second fraction diverges from the rental listing prices of the homes of the second fraction.

11. The method of claim 9 wherein the market rental rate model is a forest of classification trees each specifying a market rental rate for a combination of attributes.

12. The method of claim 11 wherein the forest of classification trees is weighted based upon its level of success in determining market rental rates for homes in the specific geographic area that were listed for rental after a specific past date, other than those used to train the model.

13. The method of claim 9, further comprising:

smoothing out a plurality of market rental rate indices periodically generated for the specific geographic area, wherein the plurality of market rental rate indices include at least a first generated market rental rate index and a second generated market rental rate index, wherein the first generated market rental rate index is different from the second generated market rental rate index.

14. The method of claim 9, further comprising:

identifying a home in the specific geographic area listed for rental after the specific past date having an attribute for which no value is available; and

in response to the identifying, imputing a value of the attribute for the identified home.

15. The method of claim 14, further comprising choosing as the imputed attribute value the median value of the attribute among homes in the specific geographic area listed for rental after the specific past date.

16. The method of claim 14, further comprising choosing as the imputed attribute value the mode value of the attribute among homes in the specific geographic area listed for rental after the specific date.

17. The method of claim 14, further comprising applying to the attribute for which no value is available a model that specifies the imputed attribute value.

18. The method of claim 17 wherein the model applied to specify the imputed attribute value is a regression model.

19. The method of claim 9, further comprising:

identifying a home in the specific geographic area listed for rental after the specific past date for which no listing price is available; and

in response to the identifying, imputing a listing price for the identified home.

20. A computer-readable hardware memory having contents configured to cause a computing system to perform a method for estimating a market rental rate for a home located in a specific geographic area, the method comprising:

retrieving data for the specific geographic area, the data comprising a plurality of entries each for a home located in the specific geographic area that was listed for rental after a specific date, each entry indicating (1) a rental listing price for the home, and, (2) values for each of a plurality of attributes including (a) rental attributes, (b) property attributes, and (c) an estimated home value attribute;

training, by a computer, a forest of classification trees to estimate a market rental rate for a home in the specific geographic area using the data retrieved for the specific geographic area;

receiving attributes of a plurality of homes in the specific geographic area;

periodically generating a market rental rate index for the specific geographic area at least in part by, for each home of the plurality of homes in the specific geographic area, applying the trained forest of classification trees to the received attributes of the home to determine a market rental rate for the home;

determining a market rental rate index for the specific geographic area using the market rental rates determined for each home of the plurality of homes in the specific geographic area; and

causing the determined market rental rate index for the specific geographic area to be provided to a user to enable the user to make a determination regarding a home in the specific geographic area.

21. A computer-readable hardware memory having contents configured to cause a computing system to perform a method for estimating a market rental rate for a home located in a specific geographic area, the method comprising:

identifying listing information for each of a plurality of homes in the specific geographic area listed for rental after a specific past date, the listing information including (1) values for (a) property attributes, (b) an estimated home value attribute, (c) rental attributes, and (2) a rental listing price for each home;

training an initialized market rental rate model to estimate a market rental rate for a home in the specific geographic area using the identified listing information;

periodically generating a market rental rate index for the specific geographic area at least in part by, for each home of the plurality of homes in the specific geographic area, applying the trained market rental rate model to attributes of the home to determine a market rental rate for the home;

determining a market rental rate index for the specific geographic area using the market rental rates estimated for each home of the plurality of homes in the specific geographic area; and

causing the determined market rental rate index for the specific geographic area to be provided to a user to enable the user to make a determination regarding a home in the specific geographic area.