Systems and Methods of Determining Vehicle Reparability
AssigneeState Farm Mutual Automobile Insurance Company
InventorsJody Ann Thoele, Jaime Skaggs, Scott T. Christensen, Ashish Sawhney, Neill Broadstone, Angela Glusick, Gustufus Phillip Theofanis
Life of this patent
Estimating the term…
Abstract
Systems and methods for determining a reparability of a vehicle are provided. In some embodiments, vehicle data is obtained, and a list of variables is generated from the obtained data. A machine learning algorithm may then be trained to generate a reparability metric by: (i) generating correlation metrics between the variables and costs to repair the vehicle, (ii) removing variables with correlation metrics below a threshold, and (iii) training the machine learning algorithm based upon unremoved variables.
Claims20
A computer-implemented method for use in determining reparability of a vehicle, the computer-implemented method comprising:
- training, by one or more processors, a machine learning algorithm to generate a reparability metric by:
- inputting variables into the machine learning algorithm;
- for each inputted variable, generating a correlation metric between the inputted variable and a cost to repair the vehicle;
- removing at least one variable from consideration by the machine learning algorithm according to a generated correlation metric corresponding to the at least one variable;
- training the machine learning algorithm based upon variables not removed from consideration by the machine learning algorithm;
- and
- further training the machine learning algorithm by re-running the machine learning algorithm;
- and
- inputting, by the one or more processors, information of a particular part into the trained machine learning algorithm to generate a reparability metric for the particular part.
A computer system for use in determining reparability of a vehicle, the system comprising:
- one or more processors;
- and
- a non-transitory program memory communicatively coupled to the one or more processors and storing executable instructions that, when executed by the one or more processors, cause the computer system to:
- train a machine learning algorithm to generate a reparability metric by:
- inputting variables into the machine learning algorithm;
- for each inputted variable, generating a correlation metric between the inputted variable and a cost to repair the vehicle;
- removing at least one variable from consideration by the machine learning algorithm according to a generated correlation metric corresponding to the at least one variable;
- training the machine learning algorithm based upon variables not removed from consideration by the machine learning algorithm;
- and
- further training the machine learning algorithm by re-running the machine learning algorithm;
- and
- input information of a particular part into the trained machine learning algorithm to generate a reparability metric for the particular part.
A non-transitory computer-readable storage medium comprising instructions that, when executed, cause one or more processors to:
- train a machine learning algorithm to generate a reparability metric by:
- inputting vehicle data including parts repair cost information, and insurance claim information into the machine learning algorithm;
- creating groups from the inputted vehicle data, the groups based upon one or more of:
- (i) frequency of repairing vehicle parts, (ii) frequency of repairing a vehicle part along with another vehicle part, (iii) cost of repairing vehicle parts, and/or (iv) insurance claims;
- training machine learning algorithm based upon the created groups;
- and
- further training the machine learning algorithm by re-running the machine learning algorithm;
- and
- input information of a particular part into the trained machine learning algorithm to generate a reparability metric for the particular part.
Dependencies are inferred from the claim text.