Predictive machine learning models
Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training and applying a machine learning model. One of the methods includes the actions of obtaining a plurality of data points associated with a parcel of real property; using a machine learning model to generate a prediction from the obtained plurality of data points, the prediction indicating a likelihood that the real property will satisfy a particular parameter, wherein the machine learning model is trained using a training set comprising a collection of data points associated with a labeled set of real property parcels distinct from the specified parcel of real property, the label indicating the particular parameter and corresponding value for each real property parcel of the training set; and based on the prediction, classifying the specified parcel of real property according to a determination of whether the predicted value of the parameter satisfies a threshold.
1. A method comprising:
obtaining, from one or more sources, a plurality of data points associated with a specified parcel of real property;
extracting specific mortgage information from the plurality of data points, comprising:
i) identifying each mortgage recorded against the parcel;
ii) identifying an indication of whether the mortgage is open; and
iii) generating a set of potentially open mortgages;
providing, as input to a machine learning model, the specific mortgage information corresponding to the set of potentially open mortgages;
adjusting accuracy of the machine learning model to minimize error in predications generated by the machine learning model based on iterative processing of a collection of data that includes data points of the specific mortgage information;
generating, by the machine learning model and after adjusting the accuracy, a prediction for each potentially open mortgage in the set of potentially open mortgages, wherein the prediction indicates a likelihood that a potentially open mortgage attached to the specified parcel of real property is actually open;
determining whether the prediction for each potentially open mortgage satisfies a threshold value; and
classifying each potentially open mortgage of the specified parcel of real property based on the determination.
2. The method of claim 1 , wherein classifying each potentially open mortgage comprises classifying the potentially open mortgage as open, closed, or requiring further analysis.
3. The method of claim 1 , further comprising determining the threshold value based on historical data of the occurrences of open mortgages resulting in title defects.
4. The method of claim 1 , further comprising determining the threshold value based on a set level of acceptable risk.
5. The method of claim 1 , further comprising:
testing an accuracy of the machining learning model against known mortgage histories for real property parcels that were not part of the training set; and
in response to determining deficiencies in the model, retraining the model using additional training data.
6. The method of claim 1 , further comprising:
determining that a change of ownership transaction occurred after a particular potentially open mortgage was recorded; and
predicting whether the particular potentially open mortgage attached to the specified parcel of real property is actually open based on the determined change of ownership transaction.
7. The method of claim 1 , further comprising:
determining that a subordinate mortgage is present without a primary mortgage;
determining that a record of the subordinate mortgage in the parcel data with no accompanying primary mortgage indicates a missing mortgage that may be open; and
predicting whether a particular potentially open mortgage attached to the specified parcel of real property is actually open based on the determined indication of the missing mortgage that may be open.
8. A system comprising:
one or more computers and one or more storage devices storing instructions that are operable, when executed by the one or more computers, to cause the one or more computers to perform operations comprising:
obtaining, from one or more sources, a plurality of data points associated with a specified parcel of real property;
extracting specific mortgage information from the plurality of data points, comprising:
i) identifying each mortgage recorded against the parcel;
ii) identifying an indication of whether the mortgage is open; and
iii) generating a set of potentially open mortgages;
providing, as input to a machine learning model, the specific mortgage information corresponding to the set of potentially open mortgages;
adjusting accuracy of the machine learning model to minimize error in predications generated by the machine learning model based on iterative processing of a collection of data that includes data points of the specific mortgage information;
generating, by the machine learning model and after adjusting the accuracy, a prediction for each potentially open mortgage in the set of potentially open mortgages, wherein the prediction indicates a likelihood that a potentially open mortgage attached to the specified parcel of real property is actually open;
determining whether the prediction for each potentially open mortgage satisfies a threshold value; and
classifying each potentially open mortgage of the specified parcel of real property based on the determination.
9. The system of claim 8 , wherein classifying each potentially open mortgage comprises classifying the potentially open mortgage as open, closed, or requiring further analysis.
10. The system of claim 8 , further comprising determining the threshold value based on historical data of the occurrences of open mortgages resulting in title defects.
11. The system of claim 8 , further comprising determining the threshold value based on a set level of acceptable risk.
12. The system of claim 8 , further comprising:
testing an accuracy of the machining learning model against known mortgage histories for real property parcels that were not part of the training set; and
in response to determining deficiencies in the model, retraining the model using additional training data.
13. The system of claim 8 , wherein the instructions are further operable to cause the one or more computers to perform operations comprising:
determining that a change of ownership transaction occurred after a particular potentially open mortgage was recorded; and
predicting whether the particular potentially open mortgage attached to the specified parcel of real property is actually open based on the determined change of ownership transaction.
14. The system of claim 8 , wherein the instructions are further operable to cause the one or more computers to perform operations comprising:
determining that a subordinate mortgage is present without a primary mortgage;
determining that a record of the subordinate mortgage in the parcel data with no accompanying primary mortgage indicates a missing mortgage that may be open; and
predicting whether a particular potentially open mortgage attached to the specified parcel of real property is actually open based on the determined indication of the missing mortgage that may be open.
15. One or more non-transitory computer-readable storage media encoded with instructions that, when executed by one or more computers, cause the one or more computers to perform operations comprising:
obtaining, from one or more sources, a plurality of data points associated with a specified parcel of real property;
extracting specific mortgage information from the plurality of data points, comprising:
i) identifying each mortgage recorded against the parcel;
ii) identifying an indication of whether the mortgage is open; and
iii) generating a set of potentially open mortgages;
providing, as input to a machine learning model, the specific mortgage information corresponding to the set of potentially open mortgages;
adjusting accuracy of the machine learning model to minimize error in predications generated by the machine learning model based on iterative processing of a collection of data that includes data points of the specific mortgage information;
generating, by the machine learning model and after adjusting the accuracy, a prediction for each potentially open mortgage in the set of potentially open mortgages, wherein the prediction indicates a likelihood that a potentially open mortgage attached to the specified parcel of real property is actually open;
determining whether the prediction for each potentially open mortgage satisfies a threshold value; and
classifying each potentially open mortgage of the specified parcel of real property based on the determination.
16. The non-transitory computer-readable storage media of claim 15 , wherein classifying each potentially open mortgage comprises classifying the potentially open mortgage as open, closed, or requiring further analysis.
17. The non-transitory computer-readable storage media of claim 15 , further comprising determining the threshold value based on historical data of the occurrences of open mortgages resulting in title defects.
18. The non-transitory computer-readable storage media of claim 15 , further comprising determining the threshold value based on a set level of acceptable risk.
19. The non-transitory computer-readable storage media of claim 15 , further comprising:
testing an accuracy of the machining learning model against known mortgage histories for real property parcels that were not part of the training set; and
in response to determining deficiencies in the model, retraining the model using additional training data.
20. The non-transitory computer-readable storage media of claim 15 , wherein the instructions, when executed, further cause the one or more computers to perform operations comprising:
determining that a change of ownership transaction occurred after a particular potentially open mortgage was recorded; and
predicting whether the particular potentially open mortgage attached to the specified parcel of real property is actually open based on the determined change of ownership transaction.
21. The non-transitory computer-readable storage media of claim 15 , wherein the instructions, when executed, further cause the one or more computers to perform operations comprising:
determining that a subordinate mortgage is present without a primary mortgage;
determining that a record of the subordinate mortgage in the parcel data with no accompanying primary mortgage indicates a missing mortgage that may be open; and
predicting whether a particular potentially open mortgage attached to the specified parcel of real property is actually open based on the determined indication of the missing mortgage that may be open.