Automated persona feature selection
Embodiments of the invention include a system for automated persona feature selection. Soft clusters of entities are received, each entity having a history of features. Each feature has a general prevalence coefficient representing prevalence of entities having the respective feature in their history. A feature list is generated for each cluster, each feature having an in-cluster coefficient representing prevalence of entities in the cluster having the feature in their history. Features having an in-cluster coefficient that is different from that feature's general prevalence coefficient are selected. A variance across the clusters is determined for each selected feature. A discriminating feature list having high variance features is generated for each cluster. Clusters are selected for an entity by comparing the features of the entity's history to features of the discriminating feature lists of the clusters. Content is customized according to the chosen clusters and sent to the entity.
1. A system comprising:
a processor;
a non-transitory computer readable storage medium storing processor-executable computer program instructions that, when executed, cause a computer processor to perform a method, the method comprising:
receiving histories of a plurality of entities, each respective history comprising a plurality of features, each entity assigned to at least one cluster from a plurality of clusters;
for each of the plurality of features corresponding to the plurality of entities, determining a corresponding general prevalence coefficient, each general prevalence coefficient representing a prevalence of entities in the plurality of entities having the respective feature in their history;
for each of the plurality of clusters, generating a respective feature list comprising features in the histories of the entities assigned to the respective cluster, each feature having a corresponding in-cluster coefficient representing a prevalence of entities in the respective cluster having the respective feature in their history;
selecting distinguishing features from the feature lists for each respective cluster by selecting features wherein a difference between each selected feature's corresponding in-cluster coefficient and corresponding general prevalence coefficient is greater than a coefficient threshold;
for each selected distinguishing feature corresponding to each respective cluster, determining a variance of the respective selected distinguishing feature's coefficients across the clusters;
generating a discriminating feature list for each respective cluster by selecting distinguishing features of the respective cluster with a respective variance greater than a variance threshold value;
choosing at least one cluster by comparing features of a history of a specified entity to features of the respective discriminating feature lists of the clusters;
customizing content according to the chosen at least one chosen cluster; and
sending the customized content to the specified entity.
2. The system of claim 1 , wherein:
the plurality of features comprises website visits.
3. The system of claim 1 , wherein:
choosing comprises choosing a plurality of clusters; and
customizing content comprises customizing content according to the plurality of chosen clusters.
4. The system of claim 1 , wherein:
at least one feature is common to at least two discriminating feature lists.
5. The system of claim 1 , wherein:
generating a discriminating feature list further comprises selecting a fixed number of distinguishing features of the respective cluster according to their respective variance.
6. The system of claim 1 , wherein:
generating a discriminating feature list further comprises selecting a fixed number of distinguishing features of the respective cluster according to their respective in-cluster prevalence.
7. A non-transitory computer readable storage medium storing processor-executable computer program instructions that, when executed, cause a computer processor to perform a method, the method comprising:
receiving histories of a plurality of entities, each respective history comprising a plurality of features, each entity assigned to at least one cluster from a plurality of clusters;
for each of the plurality of features corresponding to the plurality of entities, determining a corresponding general prevalence coefficient, each general prevalence coefficient representing a prevalence of entities in the plurality of entities having the respective feature in their history;
for each cluster, generating a respective feature list comprising features in the histories of the entities assigned to the respective cluster, each feature having a corresponding in-cluster coefficient representing prevalence of entities in the respective cluster having the respective feature in their history;
selecting distinguishing features from the feature lists for each respective cluster by selecting features wherein a difference between each selected feature's corresponding in-cluster coefficient and the selected feature's corresponding general prevalence coefficient is greater than a coefficient threshold;
for each selected distinguishing feature corresponding to each respective cluster, determining a variance of the respective selected distinguishing feature's coefficients across the clusters;
generating a discriminating feature list for each respective cluster by selecting distinguishing features of the respective cluster with a respective variance greater than a variance threshold value;
choosing at least one cluster by comparing features of a history of a specified entity to features of the respective discriminating feature lists of the clusters;
customizing content according to the at least one chosen cluster; and
sending the customized content to the specified entity.
8. The medium of claim 7 , wherein:
the plurality of features comprises website visits.
9. The medium of claim 7 , wherein:
choosing comprises choosing a plurality of clusters; and
customizing content comprises customizing content according to the plurality of chosen clusters.
10. The medium of claim 7 , wherein:
at least one feature is common to at least two discriminating feature lists.
11. A method, comprising:
receiving histories of a plurality of entities, each respective history comprising a plurality of features, each entity assigned to at least one cluster from a plurality of clusters;
for each of the plurality of features corresponding to the plurality of entities, determining a corresponding general prevalence coefficient, each general prevalence coefficient representing a prevalence of entities in the plurality of entities having the respective feature in their history;
for each cluster, generating a respective feature list comprising features in the histories of the entities assigned to the respective cluster, each feature having a corresponding in-cluster coefficient representing prevalence of entities in the respective cluster having the respective feature in their history;
selecting distinguishing features from the feature lists for each respective cluster by selecting features wherein a difference between each selected feature's corresponding in-cluster coefficient and corresponding general prevalence coefficient is greater than a coefficient threshold;
for each selected distinguishing feature corresponding to each respective cluster, determining a variance of the respective selected distinguishing feature's coefficients across the clusters;
generating a discriminating feature list for each respective cluster by selecting distinguishing features of the respective cluster with a respective variance greater than a variance threshold value;
choosing at least one cluster by comparing features of a history of a specified entity to features of the respective discriminating feature lists of the clusters;
customizing content according to the at least one chosen cluster; and
sending the customized content to the specified entity.
12. The system of claim 11 , wherein:
at least one feature is common to at least two discriminating feature lists.
13. The system of claim 1 , wherein:
each of the plurality of clusters corresponds to a persona; and
choosing at least one cluster further comprises identifying at least one persona corresponding to the specified entity by comparing features of the history of the specified entity to features of the respective discriminating feature lists of the clusters.
14. The medium of claim 7 , wherein:
each of the plurality of clusters corresponds to a persona; and
choosing at least one cluster further comprises identifying at least one persona corresponding to the specified entity by comparing features of the history of the specified entity to features of the respective discriminating feature lists of the clusters.
15. The medium of claim 7 , wherein:
generating a discriminating feature list further comprises selecting a fixed number of distinguishing features of the respective cluster according to their respective variance.
16. The medium of claim 7 , wherein:
generating a discriminating feature list further comprises selecting a fixed number of distinguishing features of the respective cluster according to their respective in-cluster prevalence.
17. The system of claim 11 , wherein:
choosing comprises choosing a plurality of clusters; and
customizing content comprises customizing content according to the plurality of chosen clusters.
18. The method of claim 11 , wherein:
the plurality of features comprises website visits.
19. The method of claim 11 , wherein:
generating a discriminating feature list further comprises selecting a fixed number of distinguishing features of the respective cluster according to their respective variance.
20. The method of claim 11 , wherein:
generating a discriminating feature list further comprises selecting a fixed number of distinguishing features of the respective cluster according to their respective in-cluster prevalence.
21. The method of claim 11 , wherein:
each of the plurality of clusters corresponds to a persona; and
choosing at least one cluster further comprises identifying at least one persona corresponding to the specified entity by comparing features of the history of the specified entity to features of the respective discriminating feature lists of the clusters.