METHODS AND APPARATUS TO GENERATE COMPUTER-TRAINED MACHINE LEARNING MODELS TO CORRECT COMPUTER-GENERATED ERRORS IN AUDIENCE DATA

1 . A system comprising:

memory;

programmable circuitry; and

machine-readable instructions, the machine-readable instructions to cause the programmable circuitry to:

aggregate first performance results of computer-generated machine learning models to generate aggregated performance results, the first performance results based on a comparison of audience member demographic data to training results, the training results generated by the computer-generated machine learning models based on at least one of: (a) a composition of a household, (b) a type of first media, (c) a daypart during which the first media was accessed, or (d) a time at which the first media was accessed;

select at least one of the computer-generated machine learning models based on a comparison of ones of the first performance results to the aggregated performance results; and

apply the at least one of the computer-generated machine learning models to correct a computer-generated error in computer-collected audience measurement data, the computer-collected audience measurement data corresponding to accesses to second media.

2 . The system of claim 1 , wherein the computer-generated error is a non-coverage error, the non-coverage error corresponding to a computer-generated measurement bias, the computer-generated measurement bias based on a computer-recordation of an impression of the second media without an association to a particular audience member that accessed the second media.

3 . The system of claim 1 , wherein the computer-generated error is a misattribution error, the misattribution error corresponding to a computer-generated measurement bias, the computer-generated measurement bias based on a recordation of an impression of the second media in association with a first demographic when the impression of the second media corresponds to a second demographic.

4 . The system of claim 1 , wherein the training results are generated by the computer-generated machine learning models during training of the computer-generated machine learning models in a privacy-protected cloud environment, the privacy-protected cloud environment being a cloud-based environment that enables multiple parties to combine their audience measurement data under privacy constraints, the privacy constraints to prevent the parties from accessing private information associated with particular audience members corresponding to the audience measurement data.

5 . The system of claim 1 , wherein the audience member demographic data corresponds to people that submit demographic information to participate in an audience member panel of an audience measurement entity.

6 . The system of claim 1 , wherein the first performance results of the computer-generated machine learning models are predictions of users that accessed the second media via client devices.

7 . The system of claim 1 , wherein the computer-collected audience measurement data includes impression data collected by a database proprietor.

8 . The system of claim 7 , wherein the database proprietor is at least one of a wireless service carrier, a social media site, or an online retailer site.

9 . A non-transitory computer readable medium comprising instructions that, when executed, cause programmable circuitry to at least:

aggregate first performance results of machine learning models to generate aggregated performance results, the first performance results based on a comparison of audience member demographic data to training results, the training results generated by the machine learning models based on inputs, the inputs to include at least one of: (a) a type of first media, (b) a daypart during which the first media was accessed, or (c) a time at which the first media was accessed;

select at least one of the machine learning models based on a comparison of ones of the first performance results to the aggregated performance results; and

apply the at least one of the machine learning models to correct a computer-generated error in audience measurement data, the audience measurement data corresponding to accesses to second media.

10 . The non-transitory computer readable medium of claim 9 , wherein the computer-generated error is a non-coverage error, the non-coverage error corresponding to a computer-generated measurement bias, the computer-generated measurement bias based on a computer-recordation of an impression of the second media without an association to a particular audience member that accessed the second media.

11 . The non-transitory computer readable medium of claim 9 , wherein the computer-generated error is a misattribution error, the misattribution error corresponding to a computer-generated measurement bias, the computer-generated measurement bias based on a recordation of an impression of the second media in association with a first demographic when the impression of the second media corresponds to a second demographic.

12 . The non-transitory computer readable medium of claim 9 , wherein the instructions are to cause the programmable circuitry to generate the training results by training the machine learning models in a privacy-protected cloud environment, the privacy-protected cloud environment being a cloud-based environment that enables multiple parties to combine their audience measurement data under privacy constraints, the privacy constraints to prevent the parties from accessing private information associated with particular audience members corresponding to the audience measurement data.

13 . The non-transitory computer readable medium of claim 9 , wherein the audience member demographic data corresponds to people that submit demographic information to participate in an audience member panel of an audience measurement entity.

14 . The non-transitory computer readable medium of claim 9 , wherein the first performance results of the machine learning models are predictions of users that accessed the second media via client devices.

15 . The non-transitory computer readable medium of claim 9 , wherein the audience measurement data includes impression data collected by a database proprietor.

16 . The non-transitory computer readable medium of claim 15 , wherein the database proprietor is at least one of a wireless service carrier, a social media site, or an online retailer site.

17 . A method comprising:

aggregating, by executing an instruction with processor circuitry, first performance results of computer-generated machine learning models to generate aggregated performance results, the first performance results based on a comparison of audience member demographic data to training results, the training results generated by the computer-generated machine learning models based on at least one of: (a) a composition of a household, (b) a daypart during which first media was accessed, or (d) a time at which the first media was accessed;

selecting, by executing an instruction with the processor circuitry, at least one of the computer-generated machine learning models based on a comparison of ones of the first performance results to the aggregated performance results; and

applying, by executing an instruction with the processor circuitry, the at least one of the computer-generated machine learning models to correct a computer-generated error in audience measurement data, the audience measurement data corresponding to accesses to second media.

18 . The method of claim 17 , wherein the computer-generated error is a non-coverage error, the non-coverage error corresponding to a computer-generated measurement bias, the computer-generated measurement bias based on a computer-recordation of an impression of the second media without an association to a particular audience member that accessed the second media.

19 . The method of claim 17 , wherein the computer-generated error is a misattribution error, the misattribution error corresponding to a computer-generated measurement bias, the computer-generated measurement bias based on a recordation of an impression of the second media in association with a first demographic when the impression of the second media corresponds to a second demographic.

20 . The method of claim 17 , further including generating the training results by training the computer-generated machine learning models in a privacy-protected cloud environment, the privacy-protected cloud environment being a cloud-based environment that enables multiple parties to combine their audience measurement data under privacy constraints, the privacy constraints to prevent the parties from accessing private information associated with particular audience members corresponding to the audience measurement data.

21 . The method of claim 17 , wherein the audience member demographic data corresponds to people that submit demographic information to participate in an audience member panel of an audience measurement entity.

22 . The method of claim 17 , wherein the first performance results of the computer-generated machine learning models are predictions of users that accessed the second media via client devices.

23 . The method of claim 17 , wherein the audience measurement data includes impression data collected by a database proprietor.

24 . The method of claim 23 , wherein the database proprietor is at least one of a wireless service carrier, a social media site, or an online retailer site.