Back to results
US 9,754,211 B2Granted
Incrementality modeling
AssigneeeBay Inc.
InventorAndes Ozbay
Life of this patent
Estimating the term…
Abstract
A decision tree methodology based uplift modeling approach, wherein one decision tree is developed simultaneously on test and control groups with the objective of maximizing the mean-difference in identified subsets within the data. The incrementality modeling approach discussed may be applied in any situation where an incrementality model is required to determine who among a group of users should be exposed to a treatment with the goal of receiving a positive response.
Claims18
A system comprising:
- a memory;
- and
- at least one hardware processor coupled to the memory and comprising instructions that cause the system to perform operations comprising:
- collecting test group data representing users who have received a treatment, wherein the test group data includes a test output value, and a set of identification variables;
- collecting control group data representing users who have not received the treatment, wherein the control group data includes a control output value, and the set of identification variables;
- dividing the test group data and the control group data into one or more microsegments based on a first identification variable from among the set of identification variables, the one or more microsegments including at least a first test microsegment and a first control microsegment, wherein the first test microsegment and the first control microsegment are defined based on the first identification variable;
- calculating a first mean difference between the first test microsegment and the first control microsegment;
- determining that the first mean difference between the first test microsegments and the first control microsegment does not transgress a threshold mean value;
- applying a second identification variable to the first test microsegment and the first control microsegment in response to the determining that the first mean difference does not transgress the threshold mean value;
- calculating a second mean difference between the first test microsegment and the first control microsegment;
- determining that the second mean difference transgresses the threshold mean value;
- creating a list that includes the first identification variable and the second identification variable in response to the determining the second mean difference transgresses the threshold mean value;
- selecting one or more users having both the first identification variable and the second identification variable, based on the list;
- and
- delivering new content to the one or more users.
12 more elements
A method comprising:
- obtaining test group data representing users who have received a treatment, wherein the test group data includes a test output value, and a set of identification variables;
- obtaining control group data representing users who have not received the treatment, wherein the control group data includes a control output value, and the set of identification variables;
- separating the test group data and the control group data into one or more microsegments based on a first identification variable from among the set of identification variables, the one or more microsegments including at least a first test microsegment and a first control microsegment, wherein the first test microsegment and the first control microsegment are defined based on the first identification variable;
- calculating a first mean difference between the first test microsegment and the first control microsegment;
- determining that the first mean difference between the first test microsegments and the first control microsegment does not transgress a threshold mean value;
- applying a second identification variable to the first test microsegment and the first control microsegment in response to the determining that the first mean difference does not transgress the threshold mean value;
- calculating a second mean difference between the first test microsegment and the first control microsegment;
- determining that the second mean difference transgresses the threshold mean value;
- generating a list that includes the first identification variable and the second identification variable in response to the determining the second mean difference transgresses the threshold mean value;
- selecting one or more users having both the first identification variable and the second identification variable, based on the list;
- and
- delivering new content to the one or more users.
9 more elements
A non-transitory computer-readable medium embodying instructions that, when executed by a processor, perform operations comprising
- obtaining test group data representing users who have received a treatment, wherein the test group data includes a test output value, and a set of identification variables;
- obtaining control group data representing users who have not received the treatment, wherein the control group data includes a control output value, and the set of identification variables;
- separating the test group data and the control group data into one or more microsegments based on a first identification variable from among the set of identification variables, the one or more microsegments including at least a first test microsegment and a first control microsegment, wherein the first test microsegment and the first control microsegment are defined based on the first identification variable;
- calculating a first mean difference between the first test microsegment and the first control microsegment;
- determining that the first mean difference between the first test microsegments and the first control microsegment does not transgress a threshold mean value;
- applying a second identification variable to the first test microsegment and the first control microsegment in response to the determining that the first mean difference does not transgress the threshold mean value;
- calculating a second mean difference between the first test microsegment and the first control microsegment;
- determining that the second mean difference transgresses the threshold mean value;
- generating a list that includes the first identification variable and the second identification variable in response to the determining the second mean difference transgresses the threshold mean value;
- selecting one or more users having, both the first identification variable and the second identification variable, based on the list;
- and
- delivering new content to the one or more users.
9 more elements
Dependencies are inferred from the claim text.