Automatically determining a current value for a real estate property, such as a home, that is tailored to input from a human user, such as its owner.

Abstract

A facility procuring information about a distinguished property from its owner that is usable to refine an automatic valuation of the distinguished property is described. The facility displays information about the distinguished property used in the automatic valuation of the distinguished property. The facility obtains user input from the owner adjusting at least one aspect of information about the distinguished property used in the automatic valuation of the distinguished property. The facility then displays to the owner a refined valuation of the distinguished property that is based on the adjustment of the obtained user input.

We claim: 
     
         1 . A computer-readable storage device storing contents adapted to cause a computing system to perform a method for procuring information about a subject home from its owner that is usable to refine an initial automatic estimated market value of the subject home, the method comprising:
 retrieving indications of home attributes of the subject home to be used in the determination of the initial automatic estimated market value of the subject home;   determining the initial automatic estimated market value of the subject home by applying, to one or more of the home attributes of the subject home, a home valuation model being usable to automatically determine automatic estimated market values for homes near the subject home based only on home attributes of the homes near the subject home, wherein the home valuation model is based on:
 a plurality of home attributes of recently-sold homes near the subject home, and 
 selling prices of the recently-sold homes near the subject home; 
   displaying at least a portion of the home attributes of the subject home used in the initial automatic estimated market value of the subject home;   obtaining input from the owner of the subject home adjusting at least one home attribute of the subject home used in the initial automatic estimated market value of the subject home; and   after obtaining the input from the owner, without human intervention,
 refining the initial automatic estimated market value, using the home valuation model, to determine a refined automatic estimated market value of the subject home based on:
 at least a portion of the home attributes of the subject home used in the determination of the initial automatic estimated market value of the subject home, and 
 the adjusted at least one home attribute of the subject home used in the initial automatic estimated market value of the subject home. 
 
   
     
     
         2 . The computer-readable storage device of  claim 1 , the method further comprising:
 constructing a first iteration of the home valuation model at least in part by:
 for each of a plurality of recently-sold homes in a geographic region that includes the subject home,
 identifying a plurality of home attributes of the recently-sold home, and 
 identifying a selling price of the recently-sold home, 
 
 selecting a fraction of the recently-sold homes, 
 selecting a fraction of the identified home attributes of the recently-sold homes, and 
 classifying the selected fraction of the recently-sold homes into two or more groups of one or more recently-sold homes based on the selected fraction of the identified home attributes of the recently-sold homes. 
   
     
     
         3 . The computer-readable storage device of  claim 2 , wherein constructing the first iteration of the home valuation model further comprises:
 for each of the two or more groups of one or more recently-sold homes,
 calculating a selling price aggregated from the one or more recently-sold homes in the group. 
   
     
     
         4 . The computer-readable storage device of  claim 1  wherein the adjustment of the obtained user input includes altering home attributes used in the refining the initial automatic estimated market value to determine the refined automatic estimated market value of the subject home. 
     
     
         5 . The computer-readable storage device of  claim 1 , the method further comprising:
 estimating a selling price of a second home based on loan values associated with a sale of the second home,   
       wherein at least one selling price of a recently-sold home near the subject home, on which the home valuation model used to determine the initial automatic estimated market value is based, is the estimated selling price of the second home. 
     
     
         6 . The computer-readable storage device of  claim 1 , wherein the obtained input from the owner of the subject home adjusting at least one home attribute of the subject home used in the initial automatic estimated market value of the subject home corrects a home attribute of the subject home used in the initial automatic estimated market value of the subject home. 
     
     
         7 . The computer-readable storage device of  claim 1 , wherein the obtained input from the owner of the subject home adjusting at least one home attribute of the subject home used in the initial automatic estimated market value of the subject home updates a home attribute of the subject home used in the initial automatic estimated market value of the subject home. 
     
     
         8 . The computer-readable storage device of  claim 1 , the method further comprising:
 obtaining input from the owner of the subject home adding at least one home attribute of the subject home not used in the initial automatic estimated market value of the subject home,   
       wherein the refined automatic estimated market value of the subject home is further based on a value for the added at least one home attribute of the subject home not used in the initial automatic estimated market value of the subject home. 
     
     
         9 . The computer-readable storage device of  claim 1 , the method further comprising:
 automatically determining automatic estimated market values for all homes in a geographic area at a standard frequency,   
       wherein the refined automatic estimated market value of the subject home is further based on at least one of the automatically determined valuations for all the homes in the geographic area. 
     
     
         10 . The computer-readable storage device of  claim 9 , wherein the standard frequency is at least weekly. 
     
     
         11 . The computer-readable storage device of  claim 1 , wherein a first iteration of the home valuation model is used to determine the initial automatic estimated market value of the subject home and wherein a second iteration of the home valuation model is used to determine the refined automatic estimated market value of the subject home. 
     
     
         12 . The computer-readable storage device of  claim 11 , the method further comprising:
 automatically refining the refined automatic estimated market value of the subject home to determine a second refined automatic estimated market value of the subject home at least in part by applying a third iteration of the home valuation model to home attributes of the subject home.   
     
     
         13 . The computer-readable storage device of  claim 1 , wherein automatically refining the initial automatic estimated market value to determine a refined automatic estimated market value of the subject home includes applying a linear regression model constructed from information about recent sales of homes near the subject home. 
     
     
         14 . A computing system, having a memory and a processor, for refining an automatic estimated market value of a subject home based on input from an owner of the subject home, the system comprising:
 a component configured to retrieve an indication of an initial automatic estimated market value of the subject home and indications of home attributes of the subject home used in the determination of the initial valuation of the subject home, the initial valuation of the subject home having been determined by applying, to one or more of the home attributes of the subject home, a home valuation model usable to automatically determine automatic estimated market values for homes based only on home attributes of those homes and that is based on selling prices of a plurality of homes;   a component configured to obtain input from the owner of the subject home adjusting at least one home attribute of the subject home used in the initial automatic estimated market value of the subject home; and   a component configured to, after obtaining the input from the owner of the subject home and without human intervention:
 determine a refined automatic estimated market valuation of the subject home that is based on:
 the obtained input from the owner of the subject home adjusting at least one home attribute of the subject home used in the initial automatic estimated market value of the subject home, 
 the home valuation model, and 
 at least a portion of the home attributes of the subject home used in the determination of the initial valuation of the subject home, and 
 
 present the refined automatic estimated market value of the subject home. 
   
     
     
         15 . The computing system of  claim 14 , wherein the component configured to present is configured to display the refined estimated market automatic valuation of the subject home to the owner of the subject home. 
     
     
         16 . The computing system of  claim 14 , wherein the component configured to present is configured to display the refined automatic estimated market value of the subject home to a user other than the owner of the subject home. 
     
     
         17 . The computing system of  claim 14 , wherein the home valuation model is a geographically-specific home valuation model and wherein the component configured to determine the refined automatic estimated market value of the subject home is configured to apply the geographically-specific home valuation model to one or more home attributes indicated by an external data source to be possessed by the subject home, 
       and wherein the adjusting at least one home attribute of the subject home used in the initial automatic estimated market value of the subject home includes altering one or more of the home attributes indicated by the external data source to be possessed by the subject home, and wherein the determined refined automatic estimated market value is based at least in part on applying the geographically-specific home valuation model to the altered home attributes. 
     
     
         18 . The computing system of  claim 17  wherein the geographically-specific home valuation model is a forest of classification trees each constructed from information about recent sales of homes near the subject home. 
     
     
         19 . The computing system of  claim 17  wherein the geographically-specific home valuation model is a linear regression model constructed from information about recent sales of homes near the subject home. 
     
     
         20 . The computing system of  claim 17  wherein the geographically-specific home valuation model is a hybrid model, utilizing both a forest of classification trees constructed from information about recent sales of home near the subject home and a linear regression-derived function constructed from information about recent sales of homes near the subject home. 
     
     
         21 . The computing system of  claim 20  wherein the component configured to determine the refined automatic estimated market value is configured to:
 divide by a first valuation of the subject home determined by the linear regression-derived function from one or more of the home attributes indicated by the external data source to be possessed by the subject home a second valuation of the subject home determined by the linear regression-derived function based on the altered home attributes to obtain a ratio; and 
 multiply the obtained ratio by a third valuation of the subject home determined by the forest of classification trees based on at least one of the home attributes indicated by the external data source to be possessed by the subject home. 
 
     
     
         22 . The computing system of  claim 20 , further comprising:
 a component configured to construct the linear regression-derived function at least in part by, for each of a plurality of recently-sold homes near the subject home,
 weighting information about the recently-sold home near the subject home based on the extent to which the recently-sold home and the subject home are similar to high-value homes near the subject home. 
   
     
     
         23 . The computing system of  claim 20 , further comprising:
 a component configured to construct the linear regression-derived function at least in part by, for each of a plurality of recently-sold homes near the subject home,
 weighting information about the recently-sold home near the subject home based on a degree of nearness of the recently-sold home to the subject home. 
   
     
     
         24 . The computing system of  claim 17  wherein the geographically-specific home valuation model is constrained to consider only home attributes available for alteration by the user. 
     
     
         25 . The computing system of  claim 14  wherein the home valuation model is a geographically-specific home valuation model and wherein the component configured to determine the refined automatic estimated market value applies the geographically-specific home valuation model to one or more home attributes indicated by an external data source to be possessed by the subject home, 
       and wherein the adjusting at least one home attribute of the subject home used in the initial automatic estimated market value of the subject home includes both (1) altering one or more of the home attributes indicated by the external data source to be possessed by the subject home and (2) adding a description of an improvement to the subject home, 
       the method further comprising detecting that the altering of one or more of the home attributes and the added description are both directed to adding a new feature to the subject home, 
       and wherein, in response to the detecting, the refined automatic estimated market value is determined based at least in part on applying the geographically-specific home valuation model to the altered home attributes, and is not based on the improvement description. 
     
     
         26 . The computing system of  claim 14 , further comprising:
 a component configured to determine the initial valuation of the subject home by applying to one or more of the home attributes of the subject home the home valuation model that is based on selling prices of the plurality of homes.   
     
     
         27 . A computing system, having a memory and a processor, comprising:
 means for automatically determining current values for homes in a geographic area; and   means for tailoring a valuation of a subject home based on information provided by an owner of the subject home.   
     
     
         28 . The computing system of  claim 27 , wherein the means for automatically determining current values for homes in a geographic area comprises:
 means for constructing a tree, wherein the means for constructing comprises:
 means for determining whether and how to split a node of a tree, and 
   means for scoring a constructed tree.   
     
     
         29 . A method, performed by a computing system having a memory and a processor, the method comprising:
 retrieving indications of home attributes of a subject home to be used in a determination of an initial automatic market valuation of the subject home, the initial automatic market valuation having been determined by applying, to one or more of the home attributes of the subject home, a home valuation model being usable to automatically determine automatic market valuations for homes near the subject home based only on home attributes of the homes near the subject home, wherein the home valuation model is based on:
 a plurality of home attributes of recently-sold homes near the subject home, and 
 selling prices of the recently-sold homes near the subject home; 
   displaying at least a portion of the home attributes of the subject home used in the initial automatic market valuation of the subject home;   obtaining input from an owner of the subject home adjusting at least one home attribute of the subject home used in the initial automatic market valuation of the subject home; and   after obtaining the input from the owner, without human intervention,
 refining, with the processor, the initial automatic market valuation, using the home valuation model, to determine a refined automatic market valuation of the subject home based on:
 at least a portion of the home attributes of the subject home used in the determination of the initial automatic market valuation of the subject home, and 
 the adjusted at least one home attribute of the subject home used in the initial automatic market valuation of the subject home. 
 
   
     
     
         30 . The method of  claim 29 , wherein the home valuation model is a random forest valuation model.