Property valuation model and visualization

1. A computer-implemented method of automated property value calculation, the method comprising:

one or more processors to performing the steps of:

receiving historic financial transaction data for a first number of real estate properties over a specified time frame;

receiving characteristic data regarding the first number of real estate properties, wherein the characteristic data comprises a number of defined categories;

creating features for the first number of real estate properties, wherein the features are attributes that are shared by all units of the first number of real estate properties;

receiving historic demographic data for a number of demographic parameters over the specified time frame, wherein the historic demographic data corresponds to defined geographic regions in which the first number of real estate properties are located;

extracting market sentiment regarding the first number of real estate properties from unstructured data sources;

determining tenant information from a tenant mapping related to the first number of real estate properties;

building and training a predictive valuation model based on the historic financial transaction data, the characteristic data, the features, and the historic demographic data;

calculating individual values for a second number of real estate properties with the predictive valuation model according to characteristics of the second number of real estate properties, wherein the calculation of individual values for the second number of real estate properties further considers the market sentiment and the tenant information of the first number of real estate properties;

aggregating the individual values of the second number of real estate properties;

receiving financial transaction data for the second number of real estate properties in real time as the financial transaction data for the second number of real estate properties come to market; and

calculating a net asset value of the second number of real estate properties according to the aggregated individual values and the financial transaction data for the second number of real estate properties; and

after training the predictive valuation model:

receiving new real estate financial transaction data, new real estate characteristic data, and new demographic data at specified periodic time intervals; and

retraining the predictive valuation model based on the new real estate financial transaction data, the new real estate characteristic data, and the new demographic data for each specified time interval.

2. The method of claim 1 , wherein training the predictive valuation model comprises:

grouping properties according to specified types and characteristics;

engineering features;

selecting features;

training a number of modeling algorithms according to the property grouping, engineered features, and selected features; and

validating model performance.

3. The method of claim 2 , further comprising:

calculating a number of similarity scores between real estate properties according to characteristics; and

organizing real estate properties into a number of sets according to their similarity scores.

4. The method of claim 1 , further comprising calculating a bias correction for the calculated net asset value.

5. The method of claim 1 , wherein the net asset valuation is calculated according to time series analysis and Kalman filtering.

6. The method of claim 1 , wherein receiving the characteristic data further comprises:

receiving the characteristic data from a number of different data sources, wherein the different data sources use different classification categories;

constructing a multi-layer property linking model that correlates the different data sources;

calculating a score for each layer of the multi-layer property linking model for each real estate property;

calculating a final score from a weighted average of the scores for the layers of the multi-layer property linking model; and

creating a universal property identifier for each real estate property based on the final score.

7. The method of claim 6 , wherein the final score determines if records in different data sources refer to the same real estate property.

8. The method of claim 1 , wherein receiving the characteristic data further comprises:

receiving characteristic data from a number of real estate-specific data sources;

receiving tenant data from a number of non-real estate-specific data sources, wherein a subset of the non-real estate-specific data sources comprise tenant occupancy data not included in the real estate-specific data sources;

constructing a multi-layer tenant linking model that correlates tenant occupancy data from the real estate-specific data sources and non-real estate-specific data sources;

calculating a score for each layer of the multi-layer tenant linking model for each real estate property;

calculating a final score from a weighted average of the scores for the layers of the multi-layer tenant linking model; and

determining tenant occupancy for each real estate property according to the final score.

9. A system for automated property value calculation, the system comprising:

a storage device configured to store program instructions; and

one or more processors operably connected to the storage device and configured to execute the program instructions to cause the system to:

receive historic financial transaction data for a first number of real estate properties over a specified time frame;

receive characteristic data regarding the first number of real estate properties, wherein the characteristic data comprises a number of defined categories;

create features for the first number of real estate properties, wherein the features are attributes that are shared by all units of the first number of real estate properties;

receive historic demographic data for a number of demographic parameters over the specified time frame, wherein the historic demographic data corresponds to defined geographic regions in which the first number of real estate properties are located;

extract market sentiment regarding the first number of real estate properties from unstructured data sources;

determining tenant information from a tenant mapping related to the first number of real estate properties;

build and train a predictive valuation model based on the historic financial transaction data, the characteristic data, the features, and the historic demographic data;

calculate individual values for a second number of real estate properties with the predictive valuation model according to characteristics of the second number of real estate properties, wherein the calculation of individual values for the second number of real estate properties further considers the market sentiment and the tenant information of the first number of real estate properties;

aggregate the individual values of the second number of real estate properties;

receive financial transaction data for the second number of real estate properties in real time as the financial transaction data for the second number of real estate properties come to market;

calculate a net asset value of the second number of real estate properties according to the aggregated individual values and the financial transaction data for the second number of real estate properties; and

after training the predictive valuation model:

receive new real estate financial transaction data, new real estate characteristic data, and new demographic data at specified periodic time intervals; and

retrain the predictive valuation model based on the new real estate financial transaction data, the new real estate characteristic data, and the new demographic data for each specified time interval.

10. The system of claim 9 , wherein training the predictive valuation model comprises:

grouping properties according to specified types and characteristics;

engineering features;

selecting features;

training a number of modeling algorithms according to the property grouping, engineered features, and selected features; and

validating model performance.

11. The system of claim 10 , further comprising:

calculating a number of similarity scores between real estate properties according to characteristics; and

organizing real estate properties into a number of sets according to their similarity scores.

12. The system of claim 9 , further comprising calculating a bias correction for the calculated net asset value.

13. The system of claim 9 , wherein the net asset valuation is calculated according to time series analysis and Kalman filtering.

14. The system of claim 9 , wherein receiving the characteristic data further comprises:

receiving the characteristic data from a number of different data sources, wherein the different data sources use different classification categories;

constructing a multi-layer property linking model that correlates the different data sources;

calculating a score for each layer of the multi-layer property linking model for each real estate property;

calculating a final score from a weighted average of the scores for the layers of the multi-layer property linking model; and

creating a universal property identifier for each real estate property based on the final score.

15. The system of claim 14 , wherein the final score determines if records in different data sources refer to the same real estate property.

16. The system of claim 9 , wherein receiving the characteristic data further comprises:

receiving characteristic data from a number of real estate-specific data sources;

receiving tenant data from a number of non-real estate-specific data sources, wherein a subset of the non-real estate-specific data sources comprise tenant occupancy data not included in the real estate-specific data sources;

constructing a multi-layer tenant linking model that correlates tenant occupancy data from the real estate-specific data sources and non-real estate-specific data sources;

calculating a score for each layer of the multi-layer tenant linking model for each real estate property;

calculating a final score from a weighted average of the scores for the layers of the multi-layer tenant linking model; and

determining tenant occupancy for each real estate property according to the final score.

17. A computer program product for automated property value calculation, the computer program product comprising:

a non-volatile computer readable storage medium having program instructions stored thereon to perform the steps of:

receiving historic financial transaction data for a first number of real estate properties over a specified time frame;

receiving characteristic data regarding the first number of real estate properties, wherein the characteristic data comprises a number of defined categories;

creating features for the first number of real estate properties, wherein the features are attributes that are shared by all units of the first number of real estate properties;

receiving historic demographic data for a number of demographic parameters over the specified time frame, wherein the historic demographic data corresponds to defined geographic regions in which the first number of real estate properties are located;

extracting market sentiment regarding the first number of real estate properties from unstructured data sources;

determining tenant information from a tenant mapping related to the first number of real estate properties;

building and training a predictive valuation model based on the historic financial transaction data, the characteristic data, the features, and the historic demographic data;

calculating individual values for a second number of real estate properties with the predictive valuation model according to characteristics of the second number of real estate properties, wherein the calculation of individual values for the second number of real estate properties also considers the market sentiment and the tenant information of the first number of real estate properties;

aggregating the individual values of the second number of real estate properties;

receiving financial transaction data for the second number of real estate properties in real time as the financial transaction data for the second number of real estate properties come to market;

calculating a net asset value of the second number of real estate properties according to the aggregated individual values and the financial transaction data for the second number of real estate properties; and

after training the predictive valuation model:

receiving new real estate financial transaction data, new real estate characteristic data, and new demographic data at specified periodic time intervals; and

retraining the predictive valuation model based on the new real estate financial transaction data, the new real estate characteristic data, and the new demographic data for each specified time interval.

18. The method of claim 1 , further comprises:

identifying, using the tenant mapping and market sentiment, anchor tenants in commercial real estate properties from the first number of real estate properties, customer interest in tenants' business, and market sentiment about tenants' business.

19. The system of claim 9 , further comprise:

identifying, using the tenant mapping and market sentiment, anchor tenants in commercial real estate properties from the first number of real estate properties, customer interest in tenants' business, and market sentiment about tenants' business.