“Dimensionality Reduction Of Multi-Attribute Consumer Profiles” in Patent Application Approval Process (USPTO 20230148331): Patent Application

2023 MAY 25 (NewsRx) -- By a News Reporter-Staff News Editor at Insurance Daily News -- A patent application by the inventors Buzi, Omri (Austin, TX, US); Dziabiak, Joshua (Austin, TX, US); Karia, Meetesh (Austin, TX, US); Lyons, Adam (Austin, TX, US); Stephens, Matthew (Austin, TX, US), filed on November 9, 2022, was made available online on May 11, 2023, according to news reporting originating from Washington, D.C., by NewsRx correspondents.

This patent application has not been assigned to a company or institution.

The following quote was obtained by the news editors from the background information supplied by the inventors: “

“The present patent is a continuation of U.S. Pat. Application 16/999,895, titled DIMENSIONALITY REDUCTION OF MULTI-ATTRIBUTE CONSUMER PROFILES, filed 21 Aug. 2020, which is a continuation of U.S. Pat. Application 15/922,412, having the same title, filed 15 Mar. 2018, which is a continuation of U.S. Pat. Application 15/601,584, having the same title, filed 22 May 2017. The entire content of each aforementioned patent filing is hereby incorporated by reference.

“The present disclosure relates generally to computer-implemented insurance comparison applications and, more specifically, to insurance comparison applications configured to inform consumers of their insurability by reducing dimensionality of attributes in their profiles that affect insurance pricing.

“Insurance, as a product, is particularly complex. There are a variety of different types of insurance, for example, homeowner’s insurance, automotive insurance, professional liability insurance, flood insurance, business continuity insurance, and the like. Within each type of insurance, often there are a variety of different types of insurance policies, often reflecting different sets of scenarios that are covered and different amounts of compensation in the event of those scenarios. Other factors often including deductibles, whether a policy provides liability or comprehensive coverage, whether a policy provides replacement value or non-replacement value compensation, and premiums for the policy. This complexity is compounded by different insurance providers offering differing terms for similar policies.

“Different consumer (also referred to as users) attributes add further complexity. For each of these types of insurance policies, there are often a relatively large number of factors that affect pricing specific to the consumer for a given policy, typically indicative of the consumer’s risk of seeking compensation under the policy (and in some cases, sensitivity to price). For example, price of a given insurance policy may depend upon 10 or more different attributes of the consumer, and in many cases substantially more attributes of consumers. Further complexity arises from differing pricing based on these user attributes by competing insurance providers (e.g., carriers or brokers), often each providing different pricing for different sets of consumer attributes in each of the different permutations of insurance above, in many cases, with the basis for prices being opaque and confusing for consumers.

“Insurance, from the perspective of the consumer, suffers from a phenomenon referred to as the curse of dimensionality. This is an open problem in the field of computer science that arises where a selection is to be made among a large number of candidates based upon a large number of factors that combine factorially. In many cases, these types of problems tend to scale poorly with the number of items considered or the number of factors upon which decisions are made. As a result, computer systems often struggle with more complex use cases, which in the field of insurance and related products, are often more closely aligned with real-world scenarios.

“These problems are often particularly apparent in existing insurance comparison software applications for consumers. Often, these software applications fail to adequately distill information about the universe of insurance products down to a relatively low dimensional output that is useful for consumers. Generally, existing systems support the evaluation of various scenarios by which consumers navigate through several dimensions related to the insurance products, but these systems often fail to provide a sufficiently low dimensional characterization of the universe of insurance products that is actionable and relevant to the consumer seeking to understand why they are offered the prices they observe, what product is best given their attributes, or how they can change their attributes to receive better pricing.”

In addition to the background information obtained for this patent application, NewsRx journalists also obtained the inventors’ summary information for this patent application: “The following is a non-exhaustive listing of some aspects of the present techniques. These and other aspects are described in the following disclosure.

“Some aspects include a process to infer insurability scores of user, the process including: receiving, with a web server, from a user computing device, via a network, a request for an insurance comparison webpage; after receiving the request, sending, to the user computing device, instructions to present one or more webpages of a website having a plurality of user inputs configured to receive a plurality of attributes of the user; receiving, with a server, from the user computing device, the attributes of the user upon entry of the attributes into the user inputs of the website; determining, with one or more processors, an insurability score with an insurability model based on the received attributes of the user, the insurability model having an output dimension of insurability score and four or more input dimensions of insurance consumer attributes; and sending, with a server, to the user computing device, instructions to display a value indicative of the insurability score, wherein the insurability score is an aggregate estimate of insurance risk that consolidates the four or more input dimensions into a single output dimension.

“Some aspects include a tangible, non-transitory, machine-readable medium storing instructions that when executed by a data processing apparatus cause the data processing apparatus to perform operations including the above-mentioned process.

“Some aspects include a system, including: one or more processors; and memory storing instructions that when executed by the processors cause the processors to effectuate operations of the above-mentioned process.

“While the present techniques are susceptible to various modifications and alternative forms, specific embodiments thereof are shown by way of example in the drawings and will herein be described in detail. The drawings may not be to scale. It should be understood, however, that the drawings and detailed description thereto are not intended to limit the present techniques to the particular form disclosed, but to the contrary, the intention is to cover all modifications, equivalents, and alternatives falling within the spirit and scope of the present techniques as defined by the appended claims.”

The claims supplied by the inventors are:

“1. A tangible, non-transitory, machine-readable medium storing instructions that when executed by one or more processors infer insurability scores of users by effectuating operations comprising: receiving, with a web server, from a user computing device, via a network, a request for an insurance comparison webpage; after receiving the request, sending, to the user computing device, instructions to present one or more webpages of a website having a plurality of user inputs configured to receive a plurality of attributes of the user; receiving, with a server, from the user computing device, the attributes of the user upon entry of the attributes into the user inputs of the website; determining, with one or more processors, an insurability score with an insurability model based on the received attributes of the user, the insurability model having an output dimension of insurability score and four or more input dimensions of insurance consumer attributes; and sending, with a server, to the user computing device, instructions to display a value indicative of the insurability score, wherein the insurability score is an aggregate estimate of insurance risk that consolidates the four or more input dimensions into a single output dimension.

“2. The medium of claim 1, the operations comprising: obtaining a calibration data set having more than 10,000 calibration records, each calibration record including a respective quote for insurance and a respective set of consumer attributes upon which the respective quote is based; and calibrating the insurability model with the calibration data before receiving the request for the insurance comparison webpage.

“3. The medium of claim 2, wherein calibrating the insurability model comprises: training a supervised machine learning model on the calibration data.

“4. The medium of claim 3, wherein training a supervised machine learning model comprises: training a plurality of candidate supervised machine learning models on different subsets of the calibration data or with different initial model parameters; and selecting among the candidate supervised machine learning models based on an aggregate measure of fitness or error of the respective candidate machine learning models relative to at least part of the calibration data.

“5. The medium of claim 3, the operations comprising: training the insurability model with a first subset of the calibration data; and validating the trained insurability model with a second subset of the calibration data that is different from the first subset of calibration data.

“6. The medium of claim 2, wherein calibrating the insurability model comprises: determining parameters of a regression of the calibration data.

“7. The medium of claim 6, wherein forming the regression comprises: fitting a function to the calibration data, wherein the function defines a differentiable surface of insurability score in a four or higher dimensional space corresponding to the four or more input dimensions.

“8. The medium of claim 2, wherein obtaining the calibration data comprises: determining representative partial attribute sets of consumers, the partial attribute sets having attributes corresponding to a first subset of the four or more dimensions, the first subset having a plurality of the four or more dimensions; and querying the insurance pricing analytics application with test consumer profiles, each test consumer profiles being formed by, for each one of the partial attribute sets, systematically varying attributes in a second subset of the four or more dimensions, the second subset being disjoint from the first subset, wherein each calibration record includes attributes of a respective one of the test consumer profiles and a query response from by the insurance pricing analytics in response to the respective one of the test consumer profiles.

“9. The medium of claim 8, wherein determining representative partial attribute sets comprises: selecting attributes in the second subset of the four or more dimensions; querying the insurance pricing analytics application with a plurality of attributes in a first subset of the four or more dimensions combined with the selected attributes in the second subset of the four or more dimensions and clustering outputs of the insurance pricing analytics into clusters; and selecting a representative partial attribute set for each of the clusters, wherein the insurance pricing analytics application is configured to output query responses for more than 100 million different permutations of test consumer profiles.

“10. The medium of claim 1, the operations comprising: determining a group of consumers that are within a threshold distance of the user in at least one dimension of the four or more input dimensions; and comparing the insurability score of the user to a measure of central tendency of insurability scores of the group of consumers, wherein: the value indicative of the insurability score is based on the comparison.

“11. The medium of claim 10, wherein: determining the group of consumers comprises determining that at least some of the group of consumers are in a geographic area of the user; and comparing the insurability score comprises: determining a difference between the insurability score of the user and the measure of central tendency; and comparing the difference to a measure of variation of the insurability scores of the group of consumers.

“12. The medium of claim 1, the operations comprising: receiving a first attribute of the user in a first dimension of the four or more input dimensions; determining a preliminary insurability score based on the first attribute and default attributes in dimensions of the four or more dimensions other than the first dimension; and sending instructions to the user computing device to display a value indicative of the preliminary insurability score before receiving all of the plurality of attributes of the user.

“13. The medium of claim 12, wherein sending instructions to the user computing device to display a value indicative of the preliminary insurability score comprises: sending instructions that cause a script executed by the user computing device to remove elements from a document object model of a webpage and add other elements to the document object model of the webpage without reloading an entire new webpage.

“14. The medium of claim 1, wherein: the insurability score is not a price, but correlates with price across a plurality of different insurance providers and a plurality of different automotive insurance policies of the respective insurance providers; the plurality of attributes and the four or more dimensions include: a value indicative of current insurance status sufficient to indicate whether a person is insured or not insured; age of a person; a value indicative of previous traffic incidents or violations of driving laws by a person; and at least two of the following: a category of vehicle use; a number of vehicles or number of drivers; a value indicative of marital status of a person; a value indicative of credit score of a person; a value indicative of whether a residence is owned by a person; a value indicative of whether a certificate of insurance is required; a value indicative of whether home insurance is also being acquired; a value indicative of whether a vehicle is owned by a person; a value indicative of annual mileage of a vehicle drive by a person; a value indicative of gender of a person; and a value indicative of an amount of education obtained by a person.

“15. The medium of claim 1, wherein sending instructions to display a value indicative of the insurability score comprises: determining an estimated mean insurability score for a geographic area including the user; causing the mean insurability score to be translated into a first angle in an angular display space; causing the insurability score of the user to be translated into a second angle in the angular display space; instructing the user computing device to display a first icon indicative of the mean insurability score at the first angle on an arc or circle; and instructing the user computing device to display a second icon indicative of the insurability score of the user at the second angle on the arc or circle.

“16. The medium of claim 1, the operations comprising: binning two or more of the plurality of attributes into two or more bins, wherein at least some attributes in different dimensions are binned into the same one of the two or more bins; wherein sending instructions to display a value indicative of the insurability score comprises instructing the user computing device to display which attributes are binned into which of the two or more bins.

“17. The medium of claim 1, wherein: the operations comprise steps for obtaining a calibration data set; the operations comprise steps for calibrating the insurability model with the calibration data set; determining an insurability score comprises steps for determining an insurability score; and sending instructions to display a value indicative of the insurability score comprises steps for forming instructions to render a user interface indicative of the insurability score.

“18. The medium of claim 1, the operations comprising: sending a plurality of insurance options to the user computing device for presentation to the user.

“19. The medium of claim 18, wherein: each of the insurance options is associated with an address of a server of a respective insurance provider of the respective insurance option.”

There are additional claims. Please visit full patent to read further.

URL and more information on this patent application, see: Buzi, Omri; Dziabiak, Joshua; Karia, Meetesh; Lyons, Adam; Stephens, Matthew. Dimensionality Reduction Of Multi-Attribute Consumer Profiles. U.S. Patent Application Number 20230148331, filed November 9, 2022 and posted May 11, 2023. Patent URL (for desktop use only): https://ppubs.uspto.gov/pubwebapp/external.html?q=(20230148331)&db=US-PGPUB&type=ids

(Our reports deliver fact-based news of research and discoveries from around the world.)