Automatically determining market rental rate index for properties

1. A computer-readable medium that is not a transitory, propagating signal per se storing a program to cause a computing system having a processor to perform a method for estimating a rental rate for homes in a named geographic area, the method comprising:

for each particular home of substantially all of the homes in the named geographic area:

accessing a data structure containing values of home attributes of homes recently listed for rental in the named geographic area and associated listing prices;

accessing values of home attributes of the particular home;

with the processor, determining a current market rental rate for the particular home by applying a model comprising at least one tree, the model created in part by:

selecting a set of housing listings;

generating the at least one tree with a root node representing a full range from the set of housing listings of each of multiple attributes; and

adding multiple nodes to the tree by iteratively creating two or more child nodes of an existing tree node, wherein each child node of the existing tree node represents an attribute subrange of an attribute range of the existing tree node, wherein applying the model comprises:

automatically comparing values of the home attributes of the particular home to values of the home attributes of homes with a recorded listing for rental in the named geographic area that correspond to nodes in the model, to identify a particular node corresponding to a subset of the homes with a recorded listing for rental in the named geographic area; and

selecting a current market rental rate for the particular home from at least one recorded rental listing price associated with the identified node; and

attributing the selected current market rental rate to the particular home;

filtering homes in the named geographic area according to a first set of filtering criteria to create a filtered set of homes;

with the processor, applying an aggregation function to the attributed current market rental rates of the homes in the filtered set of homes to obtain an aggregation result; and

causing to be displayed a characterization of the current market rental rate of homes in the filtered set of homes that is based on the aggregation result.

2. The computer-readable medium of claim 1 wherein attributing the selected current market rental rate to corresponding ones of each of substantially all of the homes in the named geographic area comprises attributing at least one current market rental rate to a home that is not rented or offered for rent.

3. The computer-readable medium of claim 1 wherein attributing the selected current market rental rate to corresponding ones of each of substantially all of the homes in the named geographic area comprises attributing at least one current market rental rate to a home that has no past rental rate.

4. The computer-readable medium of claim 1 wherein the aggregation function is mean, median, or maximum.

5. The computer-readable medium of claim 1 , the method further comprising displaying an indication of the magnitude of change from an earlier characterization of the market rental rate of homes in the named geographic area to the characterization of the current market rental rate of homes in the named geographic area.

6. The computer-readable medium of claim 1 , the method further comprising:

retrieving aggregation results obtained for each of a plurality of sets of market rental rates, each set determined for a different date; and

applying weighted spline smoothing to the retrieved and obtained aggregation results in order to obtain the displayed characterization of the current market rental rate.

7. A computing system comprising one or more processors and a computer-readable medium that is not a transitory, propagating signal per se storing a program that, when executed by the one or more processors, causes the computing system to perform a method comprising:

with the processor, training a first model, that predicts market rental rates for houses in the named geographic area, at least in part by:

selecting a set of housing listings;

generating the at least one tree with a root node representing a full range from the set of housing listings of each of multiple attributes; and

adding multiple nodes to the tree by iteratively creating two or more child nodes of an existing tree node, wherein each child node of the existing tree node represents an attribute subrange of an attribute range of the existing tree node;

for each particular home of substantially all of the homes in the named geographic area, applying the first model to the particular home's attributes to obtain a rental rate of the particular home by:

identifying one or more nodes in the first model corresponding to values of the home attributes of the particular home; and

selecting a rental rate of the particular home using at least one rental listing price associated with the identified one or more nodes;

filtering homes in the named geographic area according to a first set of filtering criteria to create a first filtered set of homes in the named geographic area; and

applying an aggregation function to the obtained rental rates of the first filtered set of homes in the named geographic area to obtain a first overall rental rate of the first filtered set of homes in the named geographic area.

8. The system of claim 7 wherein training a first model is performed at a first time and wherein the method further comprises:

at a second time later than the first time:

training a second model that predicts market rental rates for houses in the named geographic area based on their attributes, using attributes and listing prices for homes in the named geographic area that were listed for rental in a second period ending before the second time;

for each particular home of substantially all of the homes in the named geographic area, applying the second model to the particular home's attributes to obtain a rental rate of the particular home;

filtering homes in the named geographic area according to a second set of filtering criteria to create a second filtered set of homes in the named geographic area; and

applying an aggregation function to the obtained rental rates of the second filtered set of homes in the named geographic area to obtain a second overall rental rate of the second filtered set of homes in the named geographic area; and

generating a display comparing the first and second overall rental rates, wherein the display shows (A) the magnitude or percentage of change between the first and second overall rental rates and (B) the direction of change from the first overall rental rate and the second overall rental rate.

9. The system of claim 7 wherein training a first model is performed at a first time and wherein the method further comprises:

at a second time later than the first time:

training a second model that predicts market rental rates for houses in the named geographic area based on their attributes, using attributes and listing prices for homes in the named geographic area that were listed for rental in a second period ending before the second time;

for each particular home of substantially all the homes in the named geographic area, applying the second model to the particular home's attributes to obtain a rental rate of the particular home;

filtering homes in the named geographic area according to a second set of filtering criteria to create a second filtered set of homes in the names geographic area; and

applying an aggregation function to the obtained rental rates of the second filtered set of homes in the named geographic area to obtain a second overall rental rate of the second filtered set of homes in the named geographic area; and

generating a display comparing the first and second overall rental rates, wherein the display shows the annualized percentage and direction of change from the first overall rental rate and the second overall rental rate.

10. The system of claim 7 wherein training a first model is performed at a first time and wherein the method further comprises:

at a second time later than the first time:

training a second model that predicts market rental rates for houses in the named geographic area based on their attributes, using attributes and listing prices for homes in the named geographic area that were listed for rental in a second period ending before the second time;

for each particular home of substantially all of the homes in the named geographic area, applying the second model to the particular home's attributes to obtain a rental rate of the particular home;

filtering homes in the named geographic area according to a second set of filtering criteria to create a second filtered set of homes in the named geographic area; and

applying an aggregation function to the obtained rental rates of the second filtered set of homes in the named geographic area to obtain a second overall rental rate of the second filtered set of homes in the named geographic area; and

generating a display comparing the first and second overall rental rates, wherein the generated display superimposes a numerical comparison of the first and second overall rental rates over a visual depiction of the named geographic area.

11. The system of claim 10 wherein the visual depiction is a map or an aerial photograph.

12. A method in a computer system having a processor for estimating a rental rate for homes in a named geographic area, the method comprising:

for each particular home of substantially all of the homes in the named geographic area:

accessing a data structure containing values of home attributes of homes recently listed for rental in the named geographic area and associated listing prices;

accessing values of home attributes of the particular home;

with the processor, determining a current market rental rate for the particular home by applying a model comprising at least one tree, the model created in part by:

selecting a set of housing listings;

generating the at least one tree with a root node representing a full range from the set of housing listings of each of multiple attributes; and

adding multiple nodes to the tree by iteratively creating two or more child nodes of an existing tree node, wherein each child node of the existing tree node represents an attribute subrange of an attribute range of the existing tree node, wherein applying the model comprises:

automatically comparing values of the home attributes of the particular home to values of the home attributes of homes with a recorded listing for rental in the named geographic area that correspond to nodes in the model, to identify a particular node corresponding to a subset of the homes with a recorded listing for rental in the named geographic area; and

selecting a current market rental rate for the particular home from at least one recorded rental listing price associated with the identified node; and

attributing the selected current market rental rate to the particular home;

filtering homes in the named geographic area according to a first set of filtering criteria to create a filtered set of homes;

with the processor, applying an aggregation function to the current market rental rates of the homes in the filtered set of homes to obtain an aggregation result; and

causing to be displayed a characterization of the current market rental rates of homes in the filtered set of homes that is based on the aggregation result.

13. A method in a computer system having a processor the method comprising:

for each particular home of substantially all of the homes in the named geographic area:

accessing a data structure containing values of home attributes of homes recently listed for rental in the named geographic area and associated listing prices;

accessing values of home attributes of the particular home;

determining a current market rental rate for the particular home by applying a particular model, from among a set of multiple available models, wherein the particular model is selected by determining which of the multiple available models has independent variables corresponding to home attributes that most closely match the home attributes of the particular home, wherein applying the particular model comprises:

automatically comparing values of the home attributes of the particular home to values of the home attributes of homes that correspond to nodes in the model, to identify a particular node corresponding to a current market rental rate; and

attributing the current market rental rate to the particular home:

filtering homes in the named geographic area according to a first set of filtering criteria to create a filtered set of homes;

with the processor, applying an aggregation function to the attributed current market rental rates of the homes in the filtered set of homes to obtain an aggregation result; and

causing to be displayed a characterization of the current market rental rate of homes in the filtered set of homes that is based on the aggregation result.

14. The method of claim 13 wherein the aggregation function is mean. wherein the aggregation function is mean.

15. The method of claim 13 wherein the aggregation function is median.

16. The method of claim 13 wherein the aggregation function is maximum.

17. One or more computer memories that are not a transitory, propagating signal per se, collectively storing, for a named geographic area, a home rental rate display data structure, the data structure comprising:

contents configured to cause a computing system having a display device to display information including a characterization of a current market rental rate of homes in the named geographic area,

wherein the characterization of a current market rental rate of homes in the named geographic area is generated by:

for each home of substantially all of the homes in the named geographic area:

accessing a data structure containing values of home attributes of homes recently listed for rental in the named geographic area and associated listing prices;

accessing values of home attributes of the particular home;

determining a current market rental rate for the particular home by applying a particular model, from among a set of multiple available models, wherein the particular model is selected by determining which of the multiple available models has independent variables corresponding to home attributes that most closely match the home attributes of the particular home; and

attributing the current market rental rate to the particular home:

filtering homes in the named geographic area according to a first set of filtering criteria to create a filtered set of homes; and

applying an aggregation function to current market rental rates attributed to each of substantially every home in the named geographic area.

18. A computer-readable medium that is not a transitory, propagating signal per se storing a program to cause a computing system having a processor to perform a method comprising:

for each home of substantially all of the homes in the named geographic area:

accessing a data structure containing values of home attributes of homes recently listed for rental in the named geographic area and associated listing prices;

accessing values of home attributes of the particular home;

determining a current market rental rate for the particular home by applying a particular model, from among a set of multiple available models, wherein the particular model is selected by determining which of the multiple available models has independent variables corresponding to home attributes that most closely match the home attributes of the particular home; and

attributing the current market rental rate to the particular home:

filtering homes in the named geographic area according to a first set of filtering criteria to create a filtered set of homes;

with the processor, applying an aggregation function to the attributed current market rental rates of the homes in the filtered set of homes to obtain an aggregation result; and

causing to be displayed a characterization of the current market rental rate of homes in the filtered set of homes that is based on the aggregation result.

19. The computer-readable medium of claim 18 , wherein the multiple available models comprise:

a first model that predicts rental rate based upon property attributes, rental attributes, and estimated property value;

a second model that predicts rental rate based upon property attributes and rental attributes;

a third model that predict rental rate based upon property attributes and estimated value; and

a fourth model that predicts rental rate based upon only property attributes.