Methods and apparatus to model consumer choice sourcing
View Patent ↗Methods, systems and apparatus are disclosed to model consumer choices. An example apparatus includes a multinomial logit (MNL) engine to add a set of products having respondent choice data to a base MNL model, an aggregate building engine to improve a computational efficiency of model generation by generating a number of copies of the base MNL model, each one of the number of copies of the base MNL model exhibiting an effect of an independence or irrelevant alternatives (HA) property, a sourcing modifier to proportionally affect interrelationships between dissimilar ones of the number of products in the set by inserting sourcing effect values in the aggregate model, an estimator to estimate the item utility parameters of the aggregate model based on the number of copies of the base MNL model and the respondent choice data, and a simulation engine to calculate the choice probability.
1. An apparatus to calculate a choice probability, the apparatus comprising:
a choice modeling engine to retrieve respondent choice data via virtual shopping trips, the choice modeling engine to store the respondent choice data in a respondent database;
a multinomial logit (MNL) engine, implemented by at least one processor, to add a set of products to a base MNL model, the set of products retrieved from the respondent database via a network and based on the respondent choice data associated with the virtual shopping trips;
an aggregate building engine, implemented by the at least one processor, to build an aggregate model by generating a number of copies of the base MNL model, the number of copies based on a number of corresponding ones of products in the set of products associated with the virtual shopping trips;
a sourcing modifier, implemented by the at least one processor, to proportionally modify interrelationships between dissimilar ones of the number of corresponding ones of the products in the set of products to reduce a negative effect of the dissimilar ones of the number of corresponding ones of the products have on calculating the choice probability by inserting product offset values in respective ones of the number of copies of the base MNL model, the product offset values generated by subtracting sourcing effect values in the aggregate model from item utility parameters, the product offset values associated with respective ones of products in the set of products; and
a simulation engine, implemented by the at least one processor, to calculate the choice probability for the number of corresponding ones of the products in the set of products based on the product offset values.
2. The apparatus as defined in claim 1 , further including an estimator to estimate the item utility parameters of the aggregate model, the item utility parameters based on (a) the number of copies of the base MNL model and (b) the respondent choice data associated with the virtual shopping trips.
3. The apparatus as defined in claim 2 , wherein the sourcing modifier is to generate a geometric matrix when the item utility parameters overfit the respondent choice data, the geometric matrix reducing a number of the item utility parameters to produce statistically relevant convergence during estimation.
4. The apparatus as defined in claim 3 , wherein the sourcing modifier is to identify a number of spatial dimensions to generate the geometric matrix, the number of spatial dimensions to not exceed a number of products in the set of products.
5. The apparatus as defined in claim 1 , wherein the sourcing modifier is to generate a straight matrix when a geometric matrix fails to converge with the item utility parameters.
6. The apparatus as defined in claim 5 , wherein the sourcing modifier is to add index parameter placeholders in each matrix cell of the straight matrix to allow the aggregate model to consider an effect of each product in the set of products on a first product in the set of products.
7. The apparatus as defined claim 1 , wherein the aggregate building engine is to integrate a matrix with the number of copies of the base MNL model.
8. A tangible machine-readable storage medium comprising instructions that, when executed, cause a processor to, at least:
retrieve respondent choice data via virtual shopping trips, the respondent choice data to be stored in a respondent database;
add a set of products to a base multinomial logit (MNL) model, the set of products retrieved from the respondent database via a network and based on the respondent choice data associated with the virtual shopping trips;
build an aggregate model by generating a number of copies of the base MNL model, the number of copies based on a number of corresponding ones of the products in the set of products associated with the virtual shopping trips;
proportionally modify interrelationships between dissimilar ones of the number of corresponding ones of the products in the set of products to reduce a negative effect of the dissimilar ones of the number of corresponding ones of the products on calculating a choice probability by inserting product offset values in respective ones of the number of copies of the base MNL model, the product offset values generated by subtracting sourcing effect values in the aggregate model from item utility parameters, the product offset values associated with respective ones of products in the set of products; and
calculate the choice probability for the number of corresponding ones of the products in the set of products based on the product offset values.
9. The machine-readable storage medium as defined in claim 8 , further including instructions that, when executed, cause the processor to estimate the item utility parameters of the aggregate model, the item utility parameters based on (a) the number of copies of the base MNL model and (b) the respondent choice data associated with the virtual shopping trips.
10. The machine-readable storage medium as defined in claim 9 , further including instructions that, when executed, cause the processor to generate a geometric matrix when the item utility parameters overfit the respondent choice data, the geometric matrix reducing a number of the item utility parameters to produce statistically relevant convergence during estimation.
11. The machine-readable storage medium as defined in claim 10 , further including instructions that, when executed, cause the processor to identify a number of spatial dimensions to generate the geometric matrix, the number of spatial dimensions to not exceed a number of products in the set of products.
12. The machine-readable storage medium as defined in claim 8 , further including instructions that, when executed, cause the processor to generate a straight matrix when a geometric matrix fails to converge with the item utility parameters.
13. The machine-readable storage medium as defined in claim 12 , further including instructions that, when executed, cause the processor to add index parameter placeholders in each matrix cell of the straight matrix to allow the aggregate model to consider an effect of each product in the set of products on a first product in the set of products.
14. An apparatus to calculate choice probability, the apparatus comprising:
means for retrieving respondent choice data via virtual shopping trips, the respondent choice data to be stored in a respondent database;
means for adding to add a set of products to a base multinomial logit (MNL) model, the set of products retrieved from the respondent database via a network and based on the respondent choice data associated with the virtual shopping trips;
means for building to build an aggregate model by generating a number of copies of the base MNL model, the number of copies based on a number of corresponding ones of products in the set of products associated with the virtual shopping trips;
means for modifying to proportionally modify interrelationships between dissimilar ones of the number of corresponding ones of the products in the set of products to reduce a negative effect of the dissimilar ones of the number of corresponding ones of the products have on calculating the choice probability by inserting product offset matte values in respective ones of the number of copies of the base MNL model, the product offset values generated by subtracting sourcing effect values in the aggregate model from item utility parameters, the product offset values associated with respective ones of products in the set of products; and
means for calculating to calculate the choice probability for the number of corresponding ones of the products in the set of products based on the product offset values.
15. The apparatus as defined in claim 14 , further including means for estimating to estimate the item utility parameters of the aggregate model, the item utility parameters based on (a) the number of copies of the base MNL model and (b) the respondent choice data associated with the virtual shopping trips.
16. The apparatus as defined in claim 15 , wherein the means for modifying is to generate a geometric matrix when the item utility parameters overfit the respondent choice data, the geometric matrix reducing a number of the item utility parameters to produce statistically relevant convergence during estimation.
17. The apparatus as defined in claim 16 , wherein the means for modifying is to identify a number of spatial dimensions to generate the geometric matrix, the number of spatial dimensions to not exceed a number of products in the set of products.
18. The apparatus as defined in claim 14 , wherein the means for modifying is to generate a straight matrix when a geometric matrix fails to converge with the item utility parameters.
19. The apparatus as defined in claim 18 , wherein the means for modifying is to add index parameter placeholders in each matrix cell of the straight matrix to allow the aggregate model to consider an effect of each product in the set of products on a first product in the set of products.
20. The apparatus as defined in claim 14 , wherein the means for building is to integrate a matrix with the number of copies of the base MNL model.