Multi-use artificial intelligence-based ensemble model
Systems and methods for property valuation using automated valuation models are described herein, which involve a complex series of workflows for data processing, automated valuation modeling, and error detection—each of which are capable of leveraging machine learning algorithms/techniques and can be combined and operated together in a specific manner in order to improve data availability and data quality, improve the accuracy of generated property value estimates, improve the performance of the models over time, and allow the property value estimation to be adapted to various use cases. Estimation of property value may involve an ensemble model which reconciles the outputs of different sub-models based on a use case, with the sub-models using different approaches that have their own strengths and weaknesses.
1. A computer-implemented method comprising:
obtaining input data associated with a subject property;
enriching the input data;
applying a plurality of sub-models to the enriched input data to generate a plurality of sub-model outputs, wherein the plurality of sub-model outputs comprise estimations of the subject property value generated using different approaches;
predicting an error of each of the plurality of sub-model outputs based on the estimations;
determining a use case of an ensemble model;
determining a plurality of weights, wherein each weight is associated with one of the plurality of sub-model outputs, and wherein each weight is based in part on an accuracy of the respective sub-model output for the use case; and
applying as an input to the ensemble model the plurality of sub-model outputs, the plurality of weights, and the predicted error of each of the plurality of sub-model outputs, wherein application of the inputs to the ensemble model causes the ensemble model to combine the plurality of sub-model outputs into a final estimate of subject property value based on the use case.
2. The method of claim 1 , wherein the use case is selected by a user.
3. The method of claim 1 , wherein a sub-model of the plurality of sub-models comprises a county level subject-neighbors sub-model.
4. The method of claim 1 , wherein a sub-model of the plurality of sub-models comprises an appraisal adjustment regression sub-model.
5. The method of claim 1 , wherein a sub-model of the plurality of sub-models comprises an appraisal emulation sub-model.
6. The method of claim 1 , wherein a sub-model of the plurality of sub-models comprises a property level machine learning sub-model.
7. The method of claim 1 , wherein the input data comprises a property value history for the subject property, and wherein the method further comprises:
generating the property value history for the subject property based on a property value history model.
8. The method of claim 1 , wherein the input data comprises a quality and condition score for the subject property, and wherein the method further comprises:
generating the quality and condition score for the subject property based on a quality and condition model.
9. The method of claim 1 , wherein enriching the input data comprises removing duplicate transactions associated with the subject property.
10. A system comprising:
one or more computers; and
computer storage media storing instructions that when executed by the one or more computers, cause the one or more computers to perform operations comprising:
obtaining input data associated with a subject property;
enriching the input data;
applying a plurality of sub-models to the enriched input data to generate a plurality of sub-model outputs, wherein the plurality of sub-model outputs comprise estimations of the subject property value generated using different approaches;
predicting an error of each of the plurality of sub-model outputs based on the estimations;
determining a use case of an ensemble model;
determining a plurality of weights, wherein each weight is associated with one of the plurality of sub-model outputs, and wherein each weight is based in part on an accuracy of the respective sub-model output for the use case; and
applying as an input to the ensemble model the plurality of sub-model outputs, the plurality of weights, and the predicted error of each of the plurality of sub-model outputs, wherein application of the inputs to the ensemble model causes the ensemble model to combine the plurality of sub-model outputs into a final estimate of subject property value based on the use case.
11. The system of claim 10 , wherein the use case is selected by a user.
12. The system of claim 10 , wherein a sub-model of the plurality of sub-models comprises a county level subject-neighbors sub-model.
13. The system of claim 10 , wherein a sub-model of the plurality of sub-models comprises an appraisal adjustment regression sub-model.
14. The system of claim 10 , wherein a sub-model of the plurality of sub-models comprises an appraisal emulation sub-model.
15. The system of claim 10 , wherein a sub-model of the plurality of sub-models comprises a property level machine learning sub-model.
16. Non-transitory computer storage media storing instructions that when executed by a system of one or more computers, cause the one or more computers to perform operations comprising:
obtaining input data associated with a subject property;
enriching the input data;
applying a plurality of sub-models to the enriched input data to generate a plurality of sub-model outputs, wherein the plurality of sub-model outputs comprise estimations of the subject property value generated using different approaches;
predicting an error of each of the plurality of sub-model outputs based on the estimations;
determining a use case of an ensemble model;
determining a plurality of weights, wherein each weight is associated with one of the plurality of sub-model outputs, and wherein each weight is based in part on an accuracy of the respective sub-model output for the use case; and
applying as an input to the ensemble model the plurality of sub-model outputs, the plurality of weights, and the predicted error of each of the plurality of sub-model outputs, wherein application of the inputs to the ensemble model causes the ensemble model to combine the plurality of sub-model outputs into a final estimate of subject property value based on the use case.
17. The non-transitory computer storage media of claim 16 , wherein the predicted errors of the plurality of sub-model outputs are determined based on a plurality of error models associated with the plurality of sub-models, and wherein the plurality of error models are used by a model surveillance system configured to monitor performance of the plurality of sub-models and the ensemble model.