Determining an attribute of an online user using user device data

1. A computer-implemented method for determining an attribute for an online user of a candidate computing device, the method implemented using a host computing device, the method comprising:

identifying, by the host computing device, a first set of model data including a first set of device data from a plurality of model computing devices, the first set of device data including location data and access data for the plurality of model computing devices;

identifying a plurality of categories for an attribute of a population segment, each category defining a segment of the attribute, the population segment including the candidate computing device;

training a classification model by the host computing device using the first set of model data and the plurality of categories;

receiving a device identifier from the candidate computing device, the device identifier including device data of the candidate computing device;

applying the device data of the candidate computing device to the classification model to generate an offline prediction of a category of the plurality of categories for the candidate computing device;

determining a location of the candidate computing device using an IP address of the candidate computing device;

generating an online prediction of a category of the plurality of categories for the candidate computing device using the location; and

generating a composite prediction of a category of the plurality of categories for the candidate computing device by combining the offline prediction and the online prediction.

2. The method of claim 1 , further comprising:

receiving, at the host computing device from the candidate computing device, a request for content;

selecting, by the host computing device, a content item based on the composite prediction; and

serving the content item to the candidate computing device in response to the request.

3. The method of claim 1 , wherein training the classification model includes building the classification model using a logistic function.

4. The method of claim 1 , wherein the plurality of categories includes a plurality of non-overlapping ranges of household income.

5. The method of claim 1 , wherein the plurality of categories includes a plurality of overlapping ranges of education levels.

6. The method of claim 1 , wherein identifying the first set of model data further comprises:

receiving, from the plurality of computing devices, a plurality of device identifiers; and

extracting, from the plurality of device identifiers, access data associated with the plurality of computing devices, the access data including internet activity data of each model computing device of the plurality of model computing devices.

7. The method of claim 1 , wherein the device data of the candidate computing device includes internet activity data and location data associated with the candidate computing device.

8. The method of claim 1 , wherein generating the online prediction comprises retrieving, by the host computing device from a memory, a category value and distribution percentage corresponding to the location.

9. A computer system comprising:

a processor; and

a memory storing computer-executable instructions that, when executed by the processor, cause the processor to:

identify a first set of model data including a first set of device data from a plurality of model computing devices, the first set of device data including location data and access data for the plurality of model computing devices;

identify a plurality of categories for an attribute of a population segment, each category defining a segment of the attribute, the population segment including the candidate computing device;

train a classification model using the first set of model data and the plurality of categories;

receive a device identifier from the candidate computing device, the device identifier including device data of the candidate computing device;

apply the device data to the classification model to generate an offline prediction of a category of the plurality of categories for the candidate computing device;

determine a location of the candidate computing device using an IP address of the candidate computing device;

generate an online prediction of a category of the plurality of categories for the candidate computing device using the location; and

generate a composite prediction of a category of the plurality of categories for the candidate computing device by combining the offline prediction and the online prediction.

10. The computer system of claim 9 , the computer-executable instructions further causing the processor to:

receive, from the candidate computing device, a request for content;

select a content item based on the composite prediction; and

serve the content item to the candidate computing device in response to the request.

11. The computer system of claim 9 , wherein training the classification model includes building the classification model using a logistic function.

12. The computer system of claim 9 , wherein the plurality of categories includes a plurality of non-overlapping ranges of household income.

13. The computer system of claim 9 , wherein the plurality of categories includes a plurality of overlapping ranges of education levels.

14. The computer system of claim 9 , wherein identifying the first set of model data includes:

receiving, from the plurality of computing devices, a plurality of device identifiers; and

extracting, from the plurality of device identifiers, access data associated with the plurality of computing devices, the access data including internet activity data of each model computing device of the plurality of model computing devices.

15. The computer system of claim 9 , wherein the device data of the candidate computing device includes internet activity data and location data associated with the candidate computing device.

16. The computer system of claim 9 , wherein generating the online prediction comprises retrieving, by the processor from the memory, a category value and distribution percentage corresponding to the location.

17. A non-transitory, computer readable storage medium storing instructions that, when executed by a processor, cause the processor to:

identify a first set of model data including a first set of device data from a plurality of model computing devices, the first set of device data including location data and access data for the plurality of model computing devices;

identify a plurality of categories for an attribute of a population segment, each category defining a segment of the attribute, the population segment including the candidate computing device;

train a classification model using the first set of model data and the plurality of categories;

receive a device identifier from the candidate computing device, the device identifier including device data of the candidate computing device;

apply the device data to the classification model to generate an offline prediction of a category of the plurality of categories for the candidate computing device;

determine a location of the candidate computing device using an IP address of the candidate computing device;

generate an online prediction of a category of the plurality of categories for the candidate computing device using the location; and

generate a composite prediction of a category of the plurality of categories for the candidate computing device by combining the offline prediction and the online prediction.

18. The non-transitory, computer readable medium of claim 17 , further comprising instructions that cause the processor to:

receive, from the candidate computing device, a request for content;

select a content item based on the composite prediction; and

serve the content item to the candidate computing device in response to the request.

19. The non-transitory, computer readable medium of claim 17 , wherein training the classification model includes building the classification model using a logistic function.

20. The non-transitory, computer readable medium of claim 17 , wherein the plurality of categories includes a plurality of non-overlapping ranges of household income.

21. The non-transitory, computer readable medium of claim 17 , wherein the plurality of categories includes a plurality of overlapping ranges of education levels.

22. The non-transitory, computer readable medium of claim 17 , wherein identifying the first set of model data includes:

receiving, from the plurality of computing devices, a plurality of device identifiers; and

extracting, from the plurality of device identifiers, access data associated with the plurality of computing devices, the access data including internet activity data of each model computing device of the plurality of model computing devices.

23. The non-transitory, computer readable medium of claim 17 , wherein the device data of the candidate computing device includes internet activity data and location data associated with the candidate computing device.

24. The non-transitory, computer readable medium of claim 17 , wherein generating the online prediction comprises retrieving, by the processor from a memory, a category value and distribution percentage corresponding to the location.