Methods, systems, apparatus and articles of manufacture to determine causal effects

Patent metadata

Field	Value
Publication number	US-12443971-B2
Application number	US-202318400425-A
Country	US
Kind code	B2
Filing date	Dec 29, 2023
Priority date	Jun 15, 2018
Publication date	Oct 14, 2025
Grant date	Oct 14, 2025

Abstract

Methods, systems, apparatus, and articles of manufacture to determine causal effects are disclosed herein. An example apparatus includes a weighting engine to calculate a first set of weights corresponding to a first treatment dataset, a second set of weights corresponding to a second treatment dataset, and a third set of weights corresponding to a control dataset, the weighting engine to increase an operational efficiency of the apparatus by calculating the first set of weights, second set of weights, and third set of weights independently, a weighting response engine to calculate a first weighted response for the first treatment dataset, a second weighted response for the second treatment dataset, and determine a causal effect between the first treatment dataset and the second treatment dataset based on a difference between the first weighted response and the second weighted response, and a report generator to transmit a report to an audience measurement entity.

First claim

What is claimed is:

1. An audience measurement computing system comprising:
at least one processor; and
memory having stored therein computer readable instructions that, upon execution by the at least one processor, cause the audience measurement computing system to at least:
obtain a dataset indicative of a set of individuals each associated with (i) respective covariates and (ii) respective outcomes;
identify, from amongst the obtained dataset, a first treatment dataset corresponding to first individuals who have been exposed to a first treatment of an advertisement, the first treatment dataset having first covariates;
identify, from amongst the obtained dataset, a second treatment dataset corresponding to second individuals who have been exposed to a second treatment of the advertisement, the second treatment dataset having second covariates;
identify, from amongst the obtained dataset, a control dataset corresponding to third individuals who have not been exposed to the advertisement, the control dataset having third covariates;
determine at least one covariate included in the first covariates, the second covariates, and the third covariates to balance between the first and second treatment datasets and the control dataset;
simultaneously compute, via maximum entropy, first weights for the first covariates, second weights for the second covariates, and third weights for the third covariates while constraining the first weights, the second weights, and the third weights such that a sum of the first weights applied respectively to the determined at least one covariate to balance across the first individuals equals a sum of the second weights applied respectively to the determined at least one covariate to balance across the second individuals and equals a sum of the third weights applied respectively to the determined at least one covariate to balance across the third individuals;
compute a first weighted response for the first treatment dataset based on the first weights and respective outcomes corresponding to the first weights;
compute a second weighted response for the second treatment dataset based on the second weights and respective outcomes corresponding to the second weights;
compute a third weighted response for the control dataset based on the third weights and respective outcomes corresponding to the third weights;
determining an effect of the first treatment of the advertisement based on a difference between the first weighted response and the third weighted response;
determine an effect of the second treatment of the advertisement based on a difference between the second weighted response and the third weighted response;
compare the determined effect of the second treatment of the advertisement with the determined effect of the first treatment of the advertisement; and
report an indication of the determined comparison between the determined effect of the second treatment of the advertisement and the determined effect of the first treatment of the advertisement.

2. The audience measurement computing system of claim 1 , wherein the computer readable instructions further cause, upon execution by the at least one processor, the audience measurement computing system to:
identify at least one additional covariate to balance that is included in both the first treatment dataset and the control dataset, wherein simultaneously computing the first weights, the second weights, and the third weights is further constrained by the at least one additional covariate to balance such that a sum of the first weights applied respectively to the at least one additional covariate to balance across the first individuals equals a sum of the second weights applied respectively to the at least one additional covariate to balance across the second individuals and also equals a sum of the third weights applied respectively to the at least one additional covariate to balance across the third individuals.

3. The audience measurement computing system of claim 1 , wherein the computer readable instructions further cause, upon execution by the at least one processor, the audience measurement computing system to report the indication by displaying the indication via a webpage.

4. The audience measurement computing system of claim 1 , wherein the effect of the first treatment of the advertisement is indicative of an average monetary change in purchases by the first individuals.

5. The audience measurement computing system of claim 1 , wherein the computer readable instructions further cause, upon execution by the at least one processor, the audience measurement computing system to:
responsive to simultaneously computing the first weights, the second weights, and the third weights, bypass multivariate reweighting to thereby improve performance of the audience measurement computing system by computing the first weighted response and the second weighted response based on the first weights and second weights, respectively, previously simultaneously computed via maximum entropy within a single processing clock cycle.

6. The audience measurement computing system of claim 1 , wherein simultaneously computing the first weights, the second weights, and the third weights, via maximum entropy, includes computing the first weights, the second weights, and the third weights within a single processing clock cycle.

7. The audience measurement computing system of claim 1 , wherein:
the first covariates are indicative of first ages and first genders of the first individuals;
the second covariates are indicative of second ages and second genders of the second individuals; and
the third covariates are indicative of third ages and third genders of the third individuals.

8. A non-transitory computer readable medium comprising instructions that, when executed by at least one processor of a computing system, cause the computing system to at least:
obtain a dataset indicative of a set of individuals each associated with (i) respective covariates and (ii) respective outcomes;
identify, from amongst the obtained dataset, a first treatment dataset corresponding to first individuals who have been exposed to a first treatment of an advertisement, the first treatment dataset having first covariates;
identify, from amongst the obtained dataset, a second treatment dataset corresponding to second individuals who have been exposed to a second treatment of the advertisement, the second treatment dataset having second covariates
identify, from amongst the obtained dataset, a control dataset corresponding to third individuals who have not been exposed to the advertisement, the control dataset having third covariates;
determine at least one covariate included in the first covariates, the second covariates, and the third covariates to balance between the first and second treatment datasets and the control dataset;
simultaneously compute, via maximum entropy, first weights for the first covariates, second weights for the second covariates, and third weights for the third covariates while constraining the first weights, the second weights, and the third weights such that a sum of the first weights applied respectively to the determined at least one covariate to balance across the first individuals equals a sum of the second weights applied respectively to the determined at least one covariate to balance across the second individuals and also equals a sum of the third weights applied respectively to the determined at least one covariate to balance across the third individuals;
compute a first weighted response for the first treatment dataset based on the first weights and respective outcomes corresponding to the first weights;
compute a second weighted response for the second treatment dataset

Assignees

• Nielsen Co Us Llc

Inventors

• Sheppard Michael
• Daemen Ludo
• Murphy Edward
• Spoentgen Remy

Classifications

• G06F17/15 Primary

  Correlation function computation {including computation of convolution operations (arithmetic circuits for sum of products per se, e.g. multiply-accumulators G06F7/5443; digital filters, e.g. FIR, IIR, adaptive filters H03H17/00)} · CPC title

• G06F17/18

  for evaluating statistical data {, e.g. average values, frequency distributions, probability functions, regression analysis (forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04)} · CPC title

• G06Q30/0245

  Surveys · CPC title

• G06Q30/0246

  Traffic · CPC title

• G06Q30/0244 Primary

  Optimization · CPC title