Media device on/off detection using return path data
Example methods disclosed herein include accessing common homes data for a group of common homes, the common homes data including return path data and panel meter data. Disclosed example methods also include accessing common homes data for a group of common homes, the common homes data including first return path data and corresponding panel meter data associated with respective ones of the common homes, grouping the common homes data into view segments, classifying the view segments based on whether the return path data in respective ones of the view segments has matching panel meter data to determine labeled view segments, generating features from the labeled view segments, training a machine learning algorithm based on the features, and applying second return path data to the trained machine learning algorithm to determine whether a media device associated with the second return path data is on or off.
1. A method to perform media device on/off detection using return path data, the method comprising:
accessing common homes data for a group of common homes, the common homes data including first return path data and corresponding panel meter data associated with respective ones of the common homes;
grouping the common homes data into view segments;
classifying the view segments based on whether the return path data in respective ones of the view segments has matching panel meter data to determine labeled view segments;
generating a first set of features from the labeled view segments;
training a machine learning algorithm based on the first set of features, the machine learning algorithm associated with an on/off model;
applying second return path data to the machine learning algorithm trained based on the first set of features to output an on/off determination of whether a media device associated with the second return path data is on or off;
generating a second set of features from the labeled view segments;
training the machine learning algorithm based on the second set of features; and
applying the second return path data to the machine learning algorithm trained based on the second set of features to output an updated on/off determination of whether the media device associated with the second return path data is on or off.
2. The method of claim 1 , wherein the training includes setting a classification probability threshold, the classification probability threshold used by the machine learning algorithm to predict whether a media device would have matching panel meter data, thereby classifying a media associated with the return path data as on.
3. The method of claim 2 , further including adjusting the classification probability threshold based on whether the machine learning algorithm correctly identifies a media device as on or off during the training.
4. The method of claim 1 , wherein the panel meter data is data retrieved by an audience measurement entity from the group of common homes, the panel meter data including media device viewing data.
5. The method of claim 1 , wherein the machine learning algorithm is a random forest classification algorithm.
6. The method of claim 1 , wherein respective ones of the view segments have at least a 15 minute duration.
7. The method of claim 1 , wherein the features generated from the labeled view segments include a view segment index or a view segment duration, the view segment index corresponding to where in a tuning event the view segment occurs.
8. An apparatus to perform media device on/off detection using return path data, comprising:
machine readable instructions; and
processor circuitry to execute the machine readable instructions to:
access common homes data for a group of common homes, the common homes data including first return path data and corresponding panel meter data associated with respective ones of the common homes;
group the common homes data into view segments;
classify the view segments based on whether the return path data in respective ones of the view segments has matching panel meter data to determine labeled view segments;
generate a first set of features from the labeled view segments;
train a machine learning algorithm based on the first set of features;
apply second return path data to the machine learning algorithm trained based on the first set of features to output an on/off determination of whether a media device associated with the second return path data is on or off;
generate a second set of features from the labeled view segments;
train the machine learning algorithm based on the second set of features, the machine learning algorithm associated with an on/off model; and
apply the second return path data to the machine learning algorithm trained based on the second set of features to output an updated on/off determination of whether the media device associated with the second return path data is on or off.
9. The apparatus of claim 8 , wherein the processor circuitry is to set a classification probability threshold, the classification probability threshold used by the machine learning algorithm to predict whether a media device would have matching panel meter data, thereby classifying a media associated with the return path data as on.
10. The apparatus of claim 9 , wherein the processor circuitry is to adjust the classification probability threshold based on whether the machine learning algorithm correctly identifies a media device as on or off during the training.
11. The apparatus of claim 8 , wherein the panel meter data is data retrieved by an audience measurement entity from the common homes, the panel meter data including media device viewing data.
12. The apparatus of claim 8 , wherein the machine learning algorithm is a random forest classification algorithm.
13. The apparatus of claim 8 , wherein the view segments each have at least a 15 minute duration.
14. The apparatus of claim 8 , wherein features generated from the labeled view segments include a view segment index or a view segment duration, the view segment index corresponding to where in a tuning event the view segment occurs.
15. A non-transitory computer readable storage medium comprising computer readable instructions that, when executed, cause one or more processors to, at least:
access common homes data for a group of common homes, the common homes data including first return path data and corresponding panel meter data associated with respective ones of the common homes;
group the common homes data into view segments;
classify the view segments based on whether the return path data in respective ones of the view segments has matching panel meter data to determine labeled view segments;
generate a first set of features from the labeled view segments;
train a machine learning algorithm based on the first set of features;
apply second return path data to the machine learning algorithm trained based on the first set of features to output an on/off determination of whether a media device associated with the second return path data is on or off;
generate a second set of features from the labeled view segments;
train the machine learning algorithm based on the second set of features, the machine learning algorithm associated with an on/off model; and
apply the second return path data to the machine learning algorithm trained based on the second set of features to output an updated on/off determination of whether the media device associated with the second return path data is on or off.
16. The storage medium of claim 15 , wherein the instructions, when executed, further cause the one or more processors to identify a classification probability threshold used by the machine learning algorithm to predict whether a media device would have matching panel meter data, thereby classifying a media associated with the return path data as on.
17. The storage medium of claim 16 , wherein the instructions, when executed, further cause the one or more processors to adjust the classification probability threshold based on whether the algorithm correctly identifies a media device as on or off during the training.
18. The storage medium of claim 15 , wherein the machine learning algorithm is a random forest classification algorithm.
19. The storage medium of claim 15 , wherein the instructions, when executed, further cause the one or more processors to set each of the view segments to at least a 15 minute duration.
20. The storage medium of claim 15 , wherein the instructions, when executed, further cause the one or more processors to generate features from the labeled view segments such that the features include a view segment index or a view segment duration, the view segment index corresponding to where in a tuning event the view segment occurs.