Automatic analysis of regional housing markets based on the appreciation or depreciation of individual homes

1. A method in a computing system for determining a housing index value for a subject geographic region for a subject period in time, comprising:

creating a first training set comprising one or more home sales in the subject geographic region occurring at a beginning of the subject period and corresponding home attribute values;

training a first valuation model using the created first training set;

creating a second training set comprising one or more homes sales in the subject geographic region occurring at an end of the subject period and corresponding home attribute values;

training a second valuation model using the created second training set;

for each of the homes in a set of homes comprising substantially all of the homes within the subject geographic region:

determining a set of home attribute values for the home, each corresponding to a different home attribute among a set of home attributes;

applying the trained first valuation model for the subject geographic region to the set of home attribute values to generate a first valuation of the home at the beginning of the subject period;

applying the trained second valuation model for the subject geographic region to the set of home attribute values to generate a second valuation of the home at the end of the subject period;

in response to determining that the first valuation of the home or the second valuation of the home was generated based on the set of home attribute values being different at the beginning of the subject period and at the end of the subject period: (1) updating either the created first training set or the created second training set to use the set of home attributes that are the same at both the beginning and the end of the subject period, (2) modifying the trained first valuation model or the trained second valuation model to use the updated training set, and (3) regenerating either the first valuation of the home or the second valuation of the home, respectively; and

in response to successfully generating both the first valuation and the second valuation, determining an appreciation rate for the home on the basis of the generated first valuation of the home at the beginning of the subject period and the generated second valuation of the home at the end of the subject period, and otherwise, removing the home from the set of homes or imputing an estimated value of the home for at least one of the beginning of the subject period and the end of the subject period;

combining the appreciation rates determined for a subset of the set of homes to obtain an aggregate appreciation rate for the subject period by determining a weighted average of the appreciation rates determined for the subset of the set of homes, wherein a weight for each home of the subset of homes is proportional to the estimated value of the home at the beginning of the subject period; and

combining the obtained aggregate appreciation rate for the period with a housing index value for a prior period to obtain the housing index value for the subject geographic region for the subject period.

2. The method of claim 1 , further comprising causing the obtained housing index value to be displayed as part of a comparison of a housing index value for the subject geographic region and at least one other geographic region.

3. The method of claim 2 wherein the displayed comparison includes a map covering the subject geographic region and the other geographic regions.

4. The method of claim 1 , further comprising causing the obtained housing index value to be displayed as part of a trend over time of the housing index value for the subject geographic region.

5. The method of claim 1 , further comprising:

for each of a plurality of subject time periods, performing the method to obtain the housing index value for the subject geographic region for the subject period;

receiving information identifying a home in the geographic region and two points in time;

for each of the identified points in time, selecting the subject period most proximate to the point in time; and

comparing the obtained housing index values obtained for the two selected subject periods to obtain an estimate of the appreciation of the identified home between the two identified points in time.

6. The method of claim 1 , further comprising using the obtained housing index value to forecast a future housing price.

7. The method of claim 1 , further comprising using the obtained housing index value to forecast future housing prices for the subject geographic region.

8. The method of claim 1 , further comprising using the obtained housing index value together with a price-to-rent ratio or a price-to-income ratio to establish an absolute valuation indicator for the subject geographic region.

9. The method of claim 1 wherein the set of home attribute values determined for each home is determined for a time within the subject period.

10. The method of claim 1 wherein the set of home attribute values determined for each home is determined for the same time within the subject period.

11. The method of claim 1 wherein the trained first and second valuation models share the same model design.

12. The method of claim 1 wherein the subset of the set of homes discards homes of the set for which an estimated value of the home at the beginning of the subject period and an estimated value for the home at the end of the subject period could not both be obtained.

13. The method of claim 1 , wherein the subset of homes includes a distinguished home of the set for which an estimated value at the beginning of the subject period and an estimated value at the end of the subject period could not both be obtained, and wherein imputing the estimated value of the distinguished home for at least one of the beginning of the subject period and the end of the subject period comprises applying a backup home valuation model incorporating as independent variables a proper subset of the set of home attributes.

14. The method of claim 1 , wherein the subset of homes includes a distinguished home of the set for which an estimated value at the beginning of the subject period and an estimated value at the end of the subject period could not both be obtained, and wherein imputing the estimated value of the distinguished home for at least one of the beginning of the subject period and the end of the subject period comprises determining and applying a trend of estimated valuations of the home determined for a plurality of time periods proximate to the subject time period.

15. The method of claim 1 wherein the subset of homes includes a distinguished home of the set for which an estimated value at the beginning of the subject period and an estimated value at the end of the subject period could not both be obtained, and wherein imputing the estimated value of the distinguished home for at least one of the beginning of the subject period and the end of the subject period comprises:

identifying other homes in the geographic area having home attribute values similar to those of the distinguished home;

for each of one or both of the first and second valuation models:

for each of the identified homes, applying the valuation model to the identified home to obtain an estimated value for the identified home; and

aggregating the estimated values across the identified homes to obtain an estimated value for the distinguished home for the beginning or end of the subject period.

16. The method of claim 1 wherein the subset of the set of homes omits homes of the set having one or more outlier home attribute values.

17. The method of claim 1 wherein the subset of the set of homes omits homes of the set having one or more outlier valuations.

18. The method of claim 1 wherein the subset of the set of homes omits homes of the set having one or more outlier appreciation rates.

19. A computer-readable storage medium storing instructions that, when executed by a computing system, cause the computing system to perform a process for determining a housing index value for a subject geographic region for a subject period in time, the process comprising:

creating a first training set comprising one or more home sales in the subject geographic region occurring at a beginning of the subject period and corresponding home attribute values;

training a first valuation model using the created first training set;

creating a second training set comprising one or more homes sales in the subject geographic region occurring at an end of the subject period and corresponding home attribute values;

training a second valuation model using the created second training set;

for each of the homes in a set of homes comprising substantially all of the homes within the subject geographic region:

determining a set of home attribute values for the home, each corresponding to a different home attribute among a set of home attributes;

applying the trained first valuation model for the subject geographic region to the set of home attribute values to generate a first valuation of the home at the beginning of the subject period;

applying the trained second valuation model for the subject geographic region to the set of home attribute values to generate a second valuation of the home at the end of the subject period;

in response to determining that the first valuation of the home or the second valuation of the home was generated based on the set of home attribute values being different at the beginning of the subject period and at the end of the subject period: (1) updating either the created first training set or the created second training set to use the set of home attributes that are the same at both the beginning and the end of the subject period, (2) modifying the trained first valuation model or the trained second valuation model to use the updated training set, and (3) regenerating either the first valuation of the home or the second valuation of the home, respectively; and

in response to successfully generating both the first valuation and the second valuation, determining an appreciation rate for the home on the basis of the generated first valuation of the home at the beginning of the subject period and the generated second valuation of the home at the end of the subject period, and otherwise, removing the home from the set of homes or imputing an estimated value of the home for at least one of the beginning of the subject period and the end of the subject period;

combining the appreciation rates determined for a subset of the set of homes to obtain an aggregate appreciation rate for the subject period by determining a weighted average of the appreciation rates determined for the subset of the set of homes, wherein a weight for each home of the subset of homes is proportional to the estimated value of the home at the beginning of the subject period; and

combining the obtained aggregate appreciation rate for the period with a housing index value for a prior period to obtain the housing index value for the subject geographic region for the subject period.

20. A computing system for determining a housing index value for a subject geographic region for a subject period in time, comprising:

at least one processor; and

one or more memories storing instructions that, when executed by the one or more processors, cause the computing system to perform a process for determining a housing index value for a subject geographic region for a subject period in time, the process comprising:

creating a first training set comprising one or more home sales in the subject geographic region occurring at a beginning of the subject period and corresponding home attribute values;

training a first valuation model using the created first training set;

creating a second training set comprising one or more homes sales in the subject geographic region occurring at an end of the subject period and corresponding home attribute values;

training a second valuation model using the created second training set;

for each of the homes in a set of homes comprising substantially all of the homes within the subject geographic region:

determining a set of home attribute values for the home, each corresponding to a different home attribute among a set of home attributes;

applying the trained first valuation model for the subject geographic region to the set of home attribute values to generate a first valuation of the home at the beginning of the subject period;

applying the trained second valuation model for the subject geographic region to the set of home attribute values to generate a second valuation of the home at the end of the subject period;

in response to determining that the first valuation of the home or the second valuation of the home was generated based on the set of home attribute values being different at the beginning of the subject period and at the end of the subject period: (1) updating either the created first training set or the created second training set to use the set of home attributes that are the same at both the beginning and the end of the subject period, (2) modifying the trained first valuation model or the trained second valuation model to use the updated training set, and (3) regenerating either the first valuation of the home or the second valuation of the home, respectively; and

in response to successfully generating both the first valuation and the second valuation, determining an appreciation rate for the home on the basis of the generated first valuation of the home at the beginning of the subject period and the generated second valuation of the home at the end of the subject period, and otherwise, removing the home from the set of homes or imputing an estimated value of the home for at least one of the beginning of the subject period and the end of the subject period;

combining the appreciation rates determined for a subset of the set of homes to obtain an aggregate appreciation rate for the subject period by determining a weighted average of the appreciation rates determined for the subset of the set of homes, wherein a weight for each home of the subset of homes is proportional to the estimated value of the home at the beginning of the subject period; and

combining the obtained aggregate appreciation rate for the period with a housing index value for a prior period to obtain the housing index value for the subject geographic region for the subject period.

21. The compute-readable storage medium of claim 19 ,

wherein the subset of homes includes a distinguished home of the set for which an estimated value at the beginning of the subject period and an estimated value at the end of the subject period could not both be obtained; and

wherein the process further comprises:

imputing an estimated value of the distinguished home for at least one of the beginning of the subject period and the end of the subject period by applying a backup home valuation model incorporating as independent variables a proper subset of the set of home attributes.

22. The computer-readable storage medium of claim 19 , wherein the subset of homes includes a distinguished home of the set for which an estimated value at the beginning of the subject period and an estimated value at the end of the subject period could not both be obtained; and

wherein the process further comprises:

imputing an estimated value of the distinguished home for at least one of the beginning of the subject period and the end of the subject period by determining and applying a trend of estimated valuations of the home determined for a plurality of time periods proximate to the subject time period.

23. The computing system of claim 20 ,

wherein the subset of homes includes a distinguished home of the set for which an estimated value at the beginning of the subject period and an estimated value at the end of the subject period could not both be obtained; and

wherein the process further comprises:

imputing an estimated value of the distinguished home for at least one of the beginning of the subject period and the end of the subject period by applying a backup home valuation model incorporating as independent variables a proper subset of the set of home attributes.