# US 12,182,177 B2 · Methods and systems for classifying database records by introducing time dependency into time-homogeneous probability models > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,182,177 B2 - Document ID: US12182177B2 - Status: Granted - Kind: B2 - Application number: 18/462,642 - Filed: 2023-09-07 - Published: 2023-12-28 - Granted: 2024-12-31 - Earliest claimed: 2020-08-05 - Assignees: Capital One Services, LLC - Applicants: Capital One Services, LLC - Inventors: Hao Hua Huang; Bjorn Kwok - Web page: https://patentgrep.com/patent/US12182177B2 ## CPC - G06F 16/285: {Clustering or classification} - G06F 16/13: File access structures, e.g. distributed indices (arrangements of input from, or output to, record carriers G06F3/06) - G06F 16/2477: {Temporal data queries} - G06F 16/248: Presentation of query results - G06F 17/18: for evaluating statistical data {, e.g. average values, frequency distributions, probability functions, regression analysis (forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04)} - G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks - G06N 20/00: Machine learning - G06F 16/2462: {Approximate or statistical queries} - G06N 3/084: Backpropagation, e.g. using gradient descent - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS ## Abstract Methods and systems are described herein for improving data processing efficiency of classifying user files in a database. More particularly, methods and systems are described herein for improving data processing efficiency of classifying user files in a database in which the user files have a temporal element. The methods and system described herein accomplish these improvements by introducing time dependency into time-homogeneous probability models. Once time dependency has been introduced into the time-homogeneous probability models, these models may be used to improve the data processing efficiency of classifying the user files that feature a temporal element. ## Claims ### Claim 1 A system for improving data processing efficiency of classifying user files in a database into disparate states by introducing time dependency into time-homogeneous probability models, the system comprising: cloud-based storage circuitry configured to store a user record for a user; cloud-based control circuitry configured to: receiving a request to determine respective probabilities that the user record for a user belongs to each of a plurality of user record states, wherein the plurality of user record states comprises user record states defined by a time period for which the user record has been delinquent and a user record state corresponding to a charge-off of the user record; retrieving a user record history for the user record, wherein the user record history comprises periodic status checks; determining a number of the periodic status checks in the user record history; selecting a first probability model of a plurality of probability models based on the number of the periodic status checks in the user record history corresponding to a first range, wherein the first probability model comprises a time-inhomogeneous Markov chain; and receiving a first output of the first probability model indicating the respective probabilities that the user record for the user belongs to each of the plurality of user record states; and cloud-based input/output circuitry configured to generate for display, in a user interface, a recommendation based on the first output. ### Claim 2 A method of improving data processing efficiency of classifying user files in a database into disparate states by introducing time dependency into time-homogeneous probability models, the method comprising: receiving a request to determine respective probabilities that a user record for a user belongs to each of a plurality of user record states, wherein the plurality of user record states comprises user record states defined by a time period for which the user record has been delinquent; retrieving a user record history for the user record, wherein the user record history comprises periodic status checks; determining a number of the periodic status checks in the user record history; selecting a first probability model of a plurality of probability models based on the number of the periodic status checks in the user record history corresponding to a first range, wherein the first probability model comprises a time-inhomogeneous Markov chain; receiving a first output of the first probability model indicating the respective probabilities that the user record for the user belongs to each of the plurality of user record states; and generating a recommendation based on the first output. ### Claim 3 (depends on 2) The method of claim 2, wherein the time-inhomogeneous Markov chain comprises a transition probability matrix is conditional on a statement to account for differences in behavior between pre-diabolical risk and post-diabolical risk. ### Claim 4 (depends on 2) The method of claim 2, further comprising: incrementing a current time period to generate a new time period: and reapplying the first probability to the new time period. ### Claim 5 (depends on 2) The method of claim 2, wherein the recommendation indicates a pre-diabolical and post-diabolical accuracy at a predetermined interval. ### Claim 6 (depends on 2) The method of claim 2, further comprising: selecting a second probability model of the plurality of probability models based on the number of the periodic status checks in the user record history corresponding to a second range; processing a feature input using the second probability model; receiving a second output of the second probability model indicating the respective probabilities that the user record for the user belongs to each of the plurality of user record states; and generating a recommendation based on the second output. ### Claim 7 (depends on 2) The method of claim 2, wherein the first probability model is a Markov chain. ### Claim 8 (depends on 2) The method of claim 2, wherein the first probability model uses transition probability matrices to forecast a probability of an action related to the user record. ### Claim 9 (depends on 2) The method of claim 2, wherein the first output is a probability distribution row vector. ### Claim 10 (depends on 2) The method of claim 2, wherein the first range corresponds to a number from 6 to 10. ### Claim 11 (depends on 6) The method of claim 6, wherein the second range corresponds to a number greater than 11. ### Claim 12 (depends on 2) The method of claim 2, wherein the first range is based on the number of the user record states. ### Claim 13 (depends on 2) The method of claim 2, wherein the number of the user record states is 11. ### Claim 14 (depends on 2) The method of claim 2, wherein the plurality of user record states comprises a user record state corresponding to a charge-off of the user record. ### Claim 15 A non-transitory, computer-readable media for improving data processing efficiency of classifying user files in a database into disparate states by introducing time dependency into time-homogeneous probability models, comprising instructions that, when executed by one or more processors, cause operations comprising: receiving a request to determine respective probabilities that a user record for a user belongs to each of a plurality of user record states, wherein the plurality of user record states comprises user record states defined by a time period for which the user record has been delinquent, and wherein the plurality of user record states comprises a user record state corresponding to charge-off of the user record; retrieving a user record history for the user record, wherein the user record history comprises periodic status checks; determining a number of the periodic status checks in the user record history; selecting a first probability model of a plurality of probability models based on the number of the periodic status checks in the user record history corresponding to a first range; receiving a first output of the first probability model indicating the respective probabilities that the user record for the user belongs to each of the plurality of user record states; and generating a recommendation based on the first output. ### Claim 16 (depends on 15) The non-transitory, computer-readable media of claim 15, wherein the instructions further cause operations comprising: selecting a second probability model of the plurality of probability models based on the number of the periodic status checks in the user record history corresponding to a second range; processing a feature input using the second probability model; receiving a second output of the second probability model indicating the respective probabilities that the user record for the user belongs to each of the plurality of user record states; and generating a recommendation based on the second output. ### Claim 17 (depends on 15) The non-transitory, computer-readable media of claim 15, wherein the first probability model is a Markov chain. ### Claim 18 (depends on 15) The non-transitory, computer-readable media of claim 15, wherein the first probability model uses transition probability matrices to forecast a status change related to the user record. ### Claim 19 (depends on 15) The non-transitory, computer-readable media of claim 15, wherein the first output is a probability distribution row vector. ### Claim 20 (depends on 15) The non-transitory, computer-readable media of claim 15, wherein the first range corresponds to a number from 6 to 10.