Real property monitoring systems and methods for risk determination

Abstract

Machine learning techniques for determining a risk level of a target building or other type of real property include receiving data indicative of various historical characteristics of and/or associated with real property, and/or receiving data included in historical, electronic claims pertaining to buildings/real properties, and utilizing the received data to train a machine learning or other model that identifies or discovers risk factors associated with buildings/real properties. The machine learning or other model may be applied to characteristic data associated with the target building/real property to generate risk factors and/or risk indicators of the target building/real property. The techniques may include analyzing the generated risk factors and/or risk indicators to determine a risk level of the target building/real property. The risk factors, risk indicators, and/or risk level may be used for many purposes, such as pricing, quoting, underwriting, or re-underwriting of insurance policies.

What is claimed: 
     
         1 . A method implemented on a computing system comprising one or more processors and one or more memories, the method comprising:
 training, by the one or more processors, a neural network to identify a plurality of risk factors based upon a set of historical insurance claims corresponding to real properties, the neural network including a plurality of input layers, each input layer of the plurality of input layers including a respective plurality of input parameters, each input parameter of a respective plurality of input parameters corresponding to a respective characteristic of real properties;   receiving information corresponding to a target real property, the received information including respective indications of one or more characteristics of the target real property, at least one characteristic of the one or more characteristics of the target real property corresponding to an input parameter of a respective plurality of input parameters of an input layer;   analyzing, by the one or more processors, the received information using the trained neural network to determine one or more risk factors of the plurality of risk factors associated with the target real property;   determining, by the one or more processors, the risk level of the target real property based upon the one or more risk factors associated with the target real property; and   generating, by the one or more processors, an indication of the risk level of the target real property.   
     
     
         2 . The method of  claim 1 , further comprising:
 pre-processing, by the one or more processors, the set of historical insurance claims corresponding to real properties to generate one or more labels for each historical insurance claim of the set of historical insurance claims;   generating a set of labeled insurance claims by incorporating respective one or more labels with the set of insurance claims; and   training, by the one or more processors, the neural network based upon the set of labeled historical insurance claims.   
     
     
         3 . The method of  claim 2 , further comprising:
 selecting a subset of the set of labeled historical insurance claims based upon at least one label; and   training, by the one or more processors, the neural network based upon the subset of the set of labeled historical insurance claims.   
     
     
         4 . The method of  claim 3 , wherein the at least one label is associated with at least one selected from a group of a geographical area, a location type, a real property type, a percentage of an outstanding mortgage balance verse a property value, a hazard type, an age of a real property, and a type of insurance. 
     
     
         5 . The method of  claim 1 , wherein:
 the neural network is trained based upon at least one of a group consisting of one or more static characteristics of the real properties corresponding to the set of historical insurance claims and one or more dynamic characteristics of the real properties corresponding to the set of historical insurance claims, each dynamic characteristic of the one or more dynamic characteristics being associated with a change over time; and   the received information corresponding to the target real property includes respective indications of at least one of a group consisting of one or more static characteristics of the target real property and one or more dynamic characteristics of the target real property.   
     
     
         6 . The method of  claim 1 , wherein:
 the respective plurality of input parameters of the plurality of input layers of the neural network includes one or more characteristics of applicants of the set of historical insurance claims;   at least a portion of the received information of the target real property is obtained from an application for insurance for the target real property; and   the at least a portion of the received information includes respective indications of one or more characteristics of an applicant of the insurance application.   
     
     
         7 . The method of  claim 1 , further comprising:
 training the neural network based upon additional insurance claims corresponding to real properties.   
     
     
         8 . A computer system comprising one or more processors, the computer system configured to:
 train a neural network to identify a plurality of risk factors based upon a set of historical insurance claims corresponding to real properties, the neural network including a plurality of input layers, each input layer of the plurality of input layers including a respective plurality of input parameters, each input parameter of a respective plurality of input parameters corresponding to a respective characteristic of real properties;   receive information corresponding to a target real property, the received information including respective indications of one or more characteristics of the target real property, at least one characteristic of the one or more characteristics of the target real property corresponding to an input parameter of a respective plurality of input parameters of an input layer;   analyze the received information using the trained neural network to determine one or more risk factors of the plurality of risk factors associated with the target real property;   determine the risk level of the target real property based upon the one or more risk factors associated with the target real property; and   generate an indication of the risk level of the target real property.   
     
     
         9 . The computer system of  claim 8 , the computer system further configured to:
 pre-process the set of historical insurance claims corresponding to real properties to generate one or more labels for each historical insurance claim of the set of historical insurance claims;   generate a set of labeled insurance claims by incorporating respective one or more labels with the set of insurance claims; and   train the neural network based upon the set of labeled historical insurance claims.   
     
     
         10 . The computer system of  claim 9 , the computer system further configured to:
 select a subset of the set of labeled historical insurance claims based upon at least one label; and   train the neural network based upon the subset of the set of labeled historical insurance claims.   
     
     
         11 . The computer system of  claim 10 , wherein the at least one label is associated with at least one selected from a group of a geographical area, a location type, a real property type, a percentage of an outstanding mortgage balance verse a property value, a hazard type, an age of a real property, and a type of insurance. 
     
     
         12 . The computer system of  claim 8 , wherein:
 the neural network is trained based upon at least one of a group consisting of one or more static characteristics of the real properties corresponding to the set of historical insurance claims and one or more dynamic characteristics of the real properties corresponding to the set of historical insurance claims, each dynamic characteristic of the one or more dynamic characteristics being associated with a change over time; and   the received information corresponding to the target real property includes respective indications of at least one of a group consisting of one or more static characteristics of the target real property and one or more dynamic characteristics of the target real property.   
     
     
         13 . The computer system of  claim 8 , wherein:
 the respective plurality of input parameters of the plurality of input layers of the neural network includes one or more characteristics of applicants of the set of historical insurance claims;   at least a portion of the received information corresponding to the target real property is obtained from an application for insurance for the target real property; and   the at least the portion of the received information includes respective indications of one or more characteristics of an applicant of the insurance application.   
     
     
         14 . The computer system of  claim 8 , the computer system further configured to train the neural network based upon additional insurance claims corresponding to real properties. 
     
     
         15 . One or more non-transitory computer readable media having instructions stored thereon that, when executed by one or more processors, cause the one or more processors to:
 train a neural network to identify a plurality of risk factors based upon a set of historical insurance claims corresponding to real properties, the neural network including a plurality of input layers, each input layer of the plurality of input layers including a respective plurality of input parameters, each input parameter of a respective plurality of input parameters corresponding to a respective characteristic of real properties;   receive information corresponding to a target real property, the received information including respective indications of one or more characteristics of the target real property, at least one characteristic of the one or more characteristics of the target real property corresponding to an input parameter of a respective plurality of input parameters of an input layer;   analyze the received information using the trained neural network to determine one or more risk factors of the plurality of risk factors associated with the target real property;   determine the risk level of the target real property based upon the one or more risk factors associated with the target real property; and   generate an indication of the risk level of the target real property.   
     
     
         16 . The one or more non-transitory computer readable media of  claim 15  further including instructions that, when executed by one or more processors, cause the one or more processors to:
 pre-process the set of historical insurance claims corresponding to real properties to generate one or more labels for each historical insurance claim of the set of historical insurance claims; 
 generate a set of labeled insurance claims by incorporating respective one or more labels with the set of insurance claims; and 
 train the neural network based upon the set of labeled historical insurance claims. 
 
     
     
         17 . The one or more non-transitory computer readable media of  claim 16  further including instructions that, when executed by one or more processors, cause the one or more processors to:
 select a subset of the set of labeled historical insurance claims based upon at least one label; and 
 train the neural network based upon the subset of the set of labeled historical insurance claims. 
 
     
     
         18 . The one or more non-transitory computer readable media of  claim 17 , wherein the at least one label is associated with at least one selected from a group of a geographical area, a location type, a real property type, a percentage of an outstanding mortgage balance verse a property value, a hazard type, an age of a real property, and a type of insurance. 
     
     
         19 . The one or more non-transitory computer readable media of  claim 15 , wherein:
 the neural network is trained based upon at least one of a group consisting of one or more static characteristics of the real properties corresponding to the set of historical insurance claims and one or more dynamic characteristics of the real properties corresponding to the set of historical insurance claims, each dynamic characteristic of the one or more dynamic characteristics being associated with a change over time; and   the received information corresponding to the target real property includes respective indications of at least one of a group consisting of one or more static characteristics of the target real property and one or more dynamic characteristics of the target real property.   
     
     
         20 . The one or more non-transitory computer readable media of  claim 15 , wherein:
 the respective plurality of input parameters of the plurality of input layers of the neural network includes one or more characteristics of applicants of the set of historical insurance claims;   at least a portion of the received information corresponding to the target real property is obtained from an application for insurance for the target real property; and   the at least the portion of the received information includes respective indications of one or more characteristics of an applicant of the insurance application.