MACHINE LEARNING TECHNIQUES THAT IDENTIFY ATTRIBUTION OF SMALL SIGNAL STIMULUS IN NOISY RESPONSE CHANNELS
An example apparatus includes a model generator to generate a learning model based on a correlation of stimulus data and response data, the learning model to predict user responses based on stimuli presented to the users in the channel or the sub-channel, the correlation indicative of stimuli contributing to user responses at a channel or a sub-channel level. The apparatus further includes an attribution engine to determine a media spend plan based on the learning model and a budget, the media spend plan including an allocation of the budget to stimuli corresponding to the channel or the sub-channel and a user interface to display the media spend plan to a user and update the media spend plan based on predictions of the learning model when the user adjusts the budget or allocations of the media spend plan in the user interface.
1 . An apparatus comprising:
a model generator to generate a learning model based on a correlation of stimulus data and response data, the stimulus data including stimuli presented to users on a channel or a sub-channel of the channel, the response data indicative of user responses to the stimuli presented to the users, the learning model to predict user responses based on stimuli presented to the users in the channel or the sub-channel, the correlation indicative of stimuli contributing to user responses at a channel or a sub-channel level;
an attribution engine to determine a media spend plan based on the learning model and a budget, the media spend plan including an allocation of the budget to stimuli corresponding to the channel or the sub-channel; and
a user interface to display the media spend plan to a user and update the media spend plan based on predictions of the learning model when the user adjusts the budget or allocations of the media spend plan in the user interface.
2 . The apparatus of claim 1 , wherein the model generator is to adjust the learning model by providing subsets of stimulus data to the learning model and comparing responses predicted by the learning model to actual responses included in the response data.
3 . The apparatus of claim 2 , wherein the model generator is to adjust the learning model using machine learning techniques.
4 . The apparatus of claim 1 , further including a small signal correlation engine to calculate correlation coefficients based on electronic data records that include stimulus data and response data, the correlation coefficients indicative of an amount a sub-channel contributed to a response or set of responses.
5 . The apparatus of claim 4 , wherein the model generator uses the correlation coefficients to improve sub-channel predictions of the learning model.
6 . The apparatus of claim 5 , wherein the attribution engine uses the sub-channel predictions to adjust the allocations displayed by the user interface.
7 . The apparatus of claim 1 , wherein the attribution engine determines the allocation of the budget to the stimuli corresponding to the channel or the sub-channel by calculating stimulus contribution values for the stimuli based on the learning model and applying the stimulus contribution values to the budget.
8 . An apparatus comprising:
means for generating to generate a learning model based on a correlation of stimulus data and response data, the stimulus data including stimuli presented to users on a channel or a sub-channel of the channel, the response data indicative of user responses to the stimuli presented to the users, the learning model to predict user responses based on stimuli presented to the users in the channel or the sub-channel, the correlation indicative of stimuli contributing to user responses at a channel or a sub-channel level;
means for attributing to determine a media spend plan based on the learning model and a budget, the media spend plan including an allocation of the budget to stimuli corresponding to the channel or the sub-channel; and
means for displaying the media spend plan to a user and update the media spend plan based on predictions of the learning model when the user adjusts the budget or allocations of the media spend plan in the means for displaying.
9 . The apparatus of claim 8 , wherein the means for generating is to adjust the learning model by providing subsets of stimulus data to the learning model and comparing responses predicted by the learning model to actual responses included in the response data.
10 . The apparatus of claim 9 , wherein the means for generating is to adjust the learning model using machine learning techniques.
11 . The apparatus of claim 8 , further including means for correlating to calculate correlation coefficients based on electronic data records that include stimulus data and response data, the correlation coefficients indicative of an amount a sub-channel contributed to a response or set of responses.
12 . The apparatus of claim 11 , wherein the means for generating uses the correlation coefficients to improve sub-channel predictions of the learning model.
13 . The apparatus of claim 12 , wherein the means for attributing uses the sub-channel predictions to adjust the allocations displayed by the means for displaying.
14 . The apparatus of claim 8 , wherein the means for attributing determines the allocation of the budget to the stimuli corresponding to the channel or the sub-channel by calculating stimulus contribution values for the stimuli based on the learning model and applying the stimulus contribution values to the budget.
15 . A tangible computer readable storage medium comprising instructions that, when executed, cause a machine to at least:
generate a learning model based on a correlation of stimulus data and response data, the stimulus data including stimuli presented to users on a channel or a sub-channel of the channel, the response data indicative of user responses to the stimuli presented to the users, the learning model to predict user responses based on stimuli presented to the users in the channel or the sub-channel, the correlation indicative of stimuli contributing to user responses at a channel or a sub-channel level;
determine a media spend plan based on the learning model and a budget, the media spend plan including an allocation of the budget to stimuli corresponding to the channel or the sub-channel; and
display the media spend plan to a user and update the media spend plan based on predictions of the learning model when the user adjusts the budget or allocations of the media spend plan in a user interface.
16 . The tangible computer readable storage medium of claim 15 , wherein the instructions, when executed, cause the machine to adjust the model by providing subsets of stimulus data to the learning model and comparing responses predicted by the learning model to actual responses included in the response data.
17 . The tangible computer readable storage medium of claim 16 , wherein the instructions, when executed, cause the machine to adjust the learning model using machine learning techniques.
18 . The tangible computer readable storage medium of claim 15 , wherein the instructions, when executed, further cause the machine to calculate correlation coefficients based on electronic data records that includes stimulus data and response data, the correlation coefficients indicative of an amount a sub-channel contributed to a response or set of responses.
19 . The tangible computer readable storage medium of claim 18 , wherein the instructions, when executed, cause the machine to improve the sub-channel predictions of the learning model using the correlation coefficients.
20 . The tangible computer readable storage medium of claim 15 , wherein the instructions, when executed, cause the machine to determine the allocation of the budget to the stimuli corresponding to the channel or the sub-channel by calculating stimulus contribution values for the stimuli based on the learning model and applying the stimulus contribution values to the budget.