Separation of models based on presence or absence of a feature set and selection of model based on same

• US 9,298,860 B1
• Filed: 03/15/2013
• Issued: 03/29/2016
• Est. Priority Date: 03/15/2013
• Status: Active Grant

• Current Assignee: Quantcast
• Original Assignee: Quantcast
• Inventors: Preda, Daniel Ciprian, Kassakian, Peter William
• Primary Examiner(s): Shah, Kamini S
• Assistant Examiner(s): GUILL, RUSSELL L
• Application Number:
  US13/834,417
• Time in Patent Office: 1,110 Days
• US Class Current: 1/1
• CPC Class Codes: G06F 30/20 Design optimisation, verification or simulation optimisation, verification or simulation of circuit designs G06F30/30
• CPC Class Codes: G06Q 30/0202 Market predictions or forecasting for commercial activities
• CPC Class Codes: G06Q 30/0275 Auctions

Abstract

Separate models are built to predict the likelihood of conversion based on the presence or absence of one or more features. For example, a first model may be built to predict the likelihood of conversion of a non-converter who has never visited an advertiser'"'"'s website before and a second model may be built to predict the likelihood of conversion of a non-converter who has visited an advertiser'"'"'s website before. To determine which model to apply to an entity, the consumption history of the entity is searched for the presence or absence of the one or more features used to separate the models. The entity'"'"'s consumption history is then scored based on the applicable model to determine the likelihood of conversion.

Claims

1. A computer-implemented method of building differentiated models, the method comprising:

receiving a selection of one or more features on which to build separate models based on the presence or absence of the one or more features, the one or more features comprising a non-converting visit to an advertiser'"'"'s website;

applying a first filter to separate entities into a first group having the one or more features and a second group not having the one or more features, the first group of entities having at least one non-converting visit to the advertiser'"'"'s website and the second group of entities not having at least one non-converting visit to the advertiser'"'"'s website;

for the first group;

identifying a first archetypical population of entities from the first group, the first archetypical population comprising converters having a non-converting visit to the advertiser'"'"'s website prior to conversion;

identifying a first standard population of entities from the first group, the first standard population comprising non-converters which visited the advertiser'"'"'s website; and

determining a first feature set for inclusion in a first model based on a strength of correlation between each feature in the first feature set and being in the first archetypical population as compared to the first standard population;

for the second group;

identifying a second archetypical population of entities from the second group, the second archetypical population comprising converters which converted on a first visit to the advertiser'"'"'s website, the second archetypical population different from the first archetypical population;

identifying a second standard population of entities from the second group, the second standard population comprising non-converters which did not visit the advertiser'"'"'s website, the second standard population different from the first standard population; and

determining a second feature set for inclusion in a second model based on a strength of correlation between each feature in the second feature set and being in the second archetypical population as compared to the second standard population;

accessing a media consumption history of a specified entity;

determining which of the first model or the second model is applicable to the specified entity based on the presence or absence of a non-converting visit to the advertiser'"'"'s website in a media consumption history of the specified entity;

scoring the specified entity based on the applicable model; and

bidding on an opportunity to expose the specified entity to advertising content based on a result of the scoring.

2. The method of claim 1, further comprising:

for the second group;

filtering the second group to select entities having at least one additional feature;

identifying a third archetypical population from the selected entities having the at least one additional feature; and

determining a third feature set for inclusion in a third model based on a strength of correlation between each feature in the third feature set and being in the third archetypical population as compared to the second standard population.

3. The method of claim 2, further comprising:

responsive to determining that the second model is applicable to the specified entity, determining if the third model is applicable to the specified entity based on features in the specified entity'"'"'s media consumption history; and

scoring the specified entity based on a combination of scores from the applicable models.

4. The method of claim 1 further comprising:

identifying an additional archetypical population from the first group, the additional archetypical population comprising non-converters which visited the advertiser'"'"'s website;

determining an additional feature set for inclusion in a third model based on a strength of correlation between each feature in the additional feature set and being in the additional archetypical population as compared to the second standard population.

5. The method of claim 4 further comprising:

responsive to determining that the second model is applicable to the specified entity, applying the third model to the specified entity to produce a score from the third model; and

scoring the specified entity based on a combination of scores from the second model and the third model.

6. A non-transitory computer readable storage medium storing computer program instructions for building differentiated models, the computer program instructions comprising instructions for:

receiving a selection of one or more features on which to build separate models based on the presence or absence of the one or more features, the one or more features comprising a non-converting visit to an advertiser'"'"'s website;

applying a first filter to separate entities into a first group having the one or more features and a second group not having the one or more features, the first group of entities having at least one non-converting visit to the advertiser'"'"'s website and the second group of entities not having at least one non-converting visit to the advertiser'"'"'s website;

for the first group;

identifying a first archetypical population of entities from the first group, the first archetypical population comprising converters having a non-converting visit to the advertiser'"'"'s website prior to conversion;

identifying a first standard population of entities from the first group, the first standard population comprising non-converters which visited the advertiser'"'"'s website; and

determining a first feature set for inclusion in a first model based on a strength of correlation between each feature in the first feature set and being in the first archetypical population as compared to the first standard population;

for the second group;

identifying a second archetypical population of entities from the second group, the second archetypical population comprising converters which converted on a first visit to the advertiser'"'"'s website, the second archetypical population different from the first archetypical population;

identifying a second standard population of entities from the second group, the second standard population comprising non-converters which did not visit the advertiser'"'"'s website, the second standard population different from the first standard population; and

determining a second feature set for inclusion in a second model based on a strength of correlation between each feature in the second feature set and being in the second archetypical population as compared to the second standard population;

accessing a media consumption history of a specified entity;

determining which of the first model or the second model is applicable to the specified entity based on the presence or absence of a non-converting visit to the advertiser'"'"'s website in a media consumption history of the specified entity;

scoring the specified entity based on the applicable model; and

bidding on an opportunity to expose the specified entity to advertising content based on a result of the scoring.

7. The storage medium of claim 6, wherein the computer program instructions further comprise instructions for:

for the second group;

filtering the second group to select entities having at least one additional feature;

identifying a third archetypical population from the selected entities having the at least one additional feature; and

determining a third feature set for inclusion in a third model based on a strength of correlation between each feature in the third feature set and being in the third archetypical population as compared to the second standard population.

8. The storage medium of claim 7, wherein the computer program instructions further comprise instructions for:

responsive to determining that the second model is applicable to the specified entity, determining if the third model is applicable to the specified entity based on features in the specified entity'"'"'s media consumption history; and

scoring the specified entity based on a combination of scores from the applicable models.

9. A system comprising:

a processor;

a computer readable storage medium storing processor-executable computer program instructions for building differentiated models, the computer program instructions comprising instructions for;

receiving a selection of one or more features on which to build separate models based on the presence or absence of the one or more features, the one or more features comprising a non-converting visit to an advertiser'"'"'s website;

applying a first filter to separate entities into a first group having the one or more features and a second group not having the one or more features, the first group of entities having at least one non-converting visit to the advertiser'"'"'s website and the second group of entities not having at least one non-converting visit to the advertiser'"'"'s website;

for the first group;

identifying a first archetypical population of entities from the first group, the first archetypical population comprising converters having a non-converting visit to the advertiser'"'"'s website prior to conversion;

identifying a first standard population of entities from the first group, the first standard population comprising non-converters which visited the advertiser'"'"'s website; and

determining a first feature set for inclusion in a first model based on a strength of correlation between each feature in the first feature set and being in the first archetypical population as compared to the first standard population;

for the second group;

identifying a second archetypical population of entities from the second group, the second archetypical population comprising converters which converted on a first visit to the advertiser'"'"'s website, the second archetypical population different from the first archetypical population;

identifying a second standard population of entities from the second group, the second standard population comprising non-converters which did not visit the advertiser'"'"'s website, the second standard population different from the first standard population; and

determining a second feature set for inclusion in a second model based on a strength of correlation between each feature in the second feature set and being in the second archetypical population as compared to the second standard population;

accessing a media consumption history of a specified entity;

determining which of the first model or the second model is applicable to the specified entity based on the presence or absence of a non-converting visit to the advertiser'"'"'s website in a media consumption history of the specified entity;

scoring the specified entity based on the applicable model; and

bidding on an opportunity to expose the specified entity to advertising content based on a result of the scoring.

10. The system of claim 9, wherein the computer program instructions further comprise instructions for:

for the second group;

filtering the second group to select entities having at least one additional feature;

identifying a third archetypical population from the selected entities having the at least one additional feature; and

determining a third feature set for inclusion in a third model based on a strength of correlation between each feature in the third feature set and being in the third archetypical population as compared to the second standard population.

11. The system of claim 10, wherein the computer program instructions further comprise instructions for:

responsive to determining that the second model is applicable to the specified entity, determining if the third model is applicable to the specified entity based on features in the specified entity'"'"'s media consumption history; and

scoring the specified entity based on a combination of scores from the applicable models.

12. The system of claim 9, wherein the computer program instructions further comprise instructions for:

identifying an additional archetypical population from the first group, the additional archetypical population comprising non-converters which visited the advertiser'"'"'s web site;

determining an additional feature set for inclusion in a third model based on a strength of correlation between each feature in the additional feature set and being in the additional archetypical population as compared to the second standard population.

13. The system of claim 12, wherein the computer program instructions further comprise instructions for:

responsive to determining that the second model is applicable to the specified entity, applying the third model to the specified entity to produce a score from the third model; and

scoring the specified entity based on a combination of scores from the second model and the third model.

14. The system of claim 9, wherein the computer program instructions further comprise instructions for:

identifying an additional archetypical population from the first group, the additional archetypical population comprising non-converters which visited the advertiser'"'"'s website;

determining an additional feature set for inclusion in a third model based on a strength of correlation between each feature in the additional feature set and being in the additional archetypical population as compared to the second standard population.

15. The system of claim 14, wherein the computer program instructions further comprise instructions for:

responsive to determining that the second model is applicable to the specified entity, applying the third model to the specified entity to produce a score from the third model; and

scoring the specified entity based on a combination of scores from the second model and the third model.