Methods, systems, apparatus and articles of manufacture to model eCommerce sales

1. A system to model eCommerce sales, the system comprising:

trend identifier circuitry to compute trend data, the trend data to include commerce metric difference values corresponding to product sales of a plurality of eCommerce entities, the commerce metric difference values based on first commerce metrics scraped at a first time and second commerce metrics scraped at a second time, the first commerce metrics and the second commerce metrics scraped from a plurality of websites corresponding to the plurality of eCommerce entities;

splitter circuitry to split the commerce metric difference values into (a) a first trend dataset of the commerce metric difference values corresponding to a first dataset of eCommerce cooperators, in which the eCommerce cooperators have provided to the system sales data associated with the products, and (b) a second trend dataset of the commerce metric difference values corresponding to a second dataset of eCommerce non-cooperators, in which eCommerce non-cooperators sales data associated with the products is inaccessible by the system;

machine learning engine circuitry to:

train a model to estimate sales using at least a portion of the first trend dataset;

compare an output of the model to the sales data associated with the eCommerce cooperators to update the model; and

pass the second trend dataset through the model to estimate the eCommerce non-cooperators sales corresponding to the products; and

sales allocator circuitry to estimate sales missing from collected panel sales data, the estimate of missing panel sales based on the estimated eCommerce non-cooperators sales.

2. The system of claim 1 , further including a contribution probability index calculator circuitry to calculate product contribution probability indices corresponding to the products for (a) the eCommerce cooperators and (b) the eCommerce non-cooperators, the product contribution probability indices at least partially based on outputs of the model.

3. The system of claim 2 , wherein the sales allocator circuitry is to estimate the sales missing from the collected panel sales data at least partially based on the eCommerce non-cooperators and the product contribution probability indices.

4. The system of claim 1 , further including natural language processor circuitry to:

form a first set of consumer sentiments from first consumer comments about the products, the first commerce metrics including the first consumer comments; and

form a second set of consumer sentiments from second consumer comments about the products, the second commerce metrics including the second consumer comments, the commerce metric difference values including differences between the first set of consumer sentiments and the second set of consumer sentiments.

5. The system of claim 1 , wherein the first commerce metrics include first ratings metrics for the products and first traffic metrics for the products, the second commerce metrics include second ratings metrics for the products and second traffic metrics for the products, and the commerce metric difference values include (a) differences between the first ratings metrics and the second ratings metrics, and (b) differences between the first traffic metrics and the second traffic metrics.

6. The system of claim 5 , wherein the first traffic metrics are based on a number of first consumer feedback instances for the products summed with a number of first consumer comments about the products, and wherein the second traffic metrics are based on a number of second consumer feedback instances for the products summed with a number of second consumer comments about the products.

7. The system of claim 1 , further including panel gap analyzer circuitry to identify a statistical gap in a panel composition for a first one of the products by computing a difference of a first ratio of a first number of sales of the first one of the products to a panel and a second number of sales of all of the products to the panel, and a second ratio of a third number of sales of the first one of the products recorded by eCommerce cooperators and a fourth number of sales of all of the products to the eCommerce cooperators.

8. The system of claim 7 , wherein the statistical gap is a first statistical gap in the panel composition for the first one of the products, and the panel gap analyzer circuitry is to:

identify a second statistical gap in the panel composition for a second one of the products; and

combine the first statistical gap and the second statistical gap to identify a third statistical gap in the panel composition related to at least one of a strata, or a consumer class.

9. A system to model eCommerce sales, the system comprising:

means for determining trends to compute commerce metric difference value trend data corresponding to product sales of a plurality of eCommerce entities, the commerce metric difference values based on first commerce metrics scraped at a first time and second commerce metrics scraped at a second time, the first commerce metrics and the second commerce metrics scraped from a plurality of websites corresponding to the plurality of eCommerce entities;

means for splitting to split the commerce metric difference value trend data into (a) a first trend dataset of the commerce metric difference value trend data corresponding to a first dataset of eCommerce cooperators, in which the eCommerce cooperators have provided to the system sales data associated with the products, and (b) a second trend dataset of the commerce metric difference value trend data corresponding to a second dataset of eCommerce non-cooperators, in which eCommerce non-cooperators sales data associated with the products is inaccessible by the system;

means for predicting to estimate eCommerce non-cooperators sales corresponding to the products by training a model to estimate sales using at least a portion of the first trend dataset, comparing an output of the model to the sales data associated with the eCommerce cooperators to update the model; and passing the second trend dataset through the model; and

means for allocating sales to estimate sales missing from collected panel sales data, the estimate of missing panel sales based on the estimated eCommerce non-cooperators sales.

10. The system of claim 9 , further including means for language processing to:

form a first set of consumer sentiments from first consumer comments about the products, the first commerce metrics including the first consumer comments; and

form a second set of consumer sentiments from second consumer comments about the products, the second commerce metrics including the second consumer comments, the commerce metric difference values including differences between the first set of consumer sentiments and the second set of consumer sentiments.

11. The system of claim 9 , wherein the first commerce metrics include first ratings metrics for the products and first traffic metrics for the products, the second commerce metrics include second ratings metrics for the products and second traffic metrics for the products, and the commerce metric difference values include (a) differences between the first ratings metrics and the second ratings metrics, and (b) differences between the first traffic metrics and the second traffic metrics.

12. The system of claim 9 , further including means for determining gaps to identify a statistical gap in a panel composition for a first one of the products by computing a difference of a first ratio of a first number of sales of the first one of the products to a panel and a second number of sales of all of the products to the panel, and a second ratio of a third number of sales of the first one of the products recorded by eCommerce cooperators and a fourth number of sales of all of the products to the eCommerce cooperators.

13. The system of claim 12 , wherein the statistical gap is a first statistical gap in the panel composition for the first one of the products, and the determining gaps means is to:

identify a second statistical gap in the panel composition for a second one of the products; and

combine the first statistical gap and the second statistical gap to identify a third statistical gap in the panel composition related to at least one of a strata, or a consumer class.

14. A non-transitory computer-readable storage medium comprising instructions that, when executed, cause a machine to:

calculate trend data, including commerce metric difference values corresponding to product sales of a plurality of eCommerce entities, the commerce metric difference values based on first commerce metrics scraped at a first time and second commerce metrics scraped at a second time, the first commerce metrics and the second commerce metrics scraped from a plurality of websites corresponding to the plurality of eCommerce entities;

split the commerce metric difference value trend data into (a) a first trend dataset of the commerce metric difference values corresponding to a first dataset of eCommerce cooperators, in which the eCommerce cooperators have provided to the machine to collect sales data associated with the products, and (b) a second trend dataset of the commerce metric difference values corresponding to a second dataset of eCommerce non-cooperators, in which eCommerce non-cooperators sales data associated with the products is inaccessible by the machine;

estimate eCommerce non-cooperators sales corresponding to the products by training a model to estimate male using at least a portion of the first trend dataset, comparing an output of the model to the sales data associated with the eCommerce cooperators to update the model, and running the second trend dataset through the model; and

estimate sales missing from collected panel sales data, the estimate of missing panel sales based on the estimated eCommerce non-cooperators sales.

15. The non-transitory computer-readable storage medium of claim 14 , wherein the instructions, when executed, cause the machine to:

apply natural language processing to first consumer comments about the products to form a first set of consumer sentiments, the first commerce metrics including the first consumer comments; and

apply the natural language processing to second consumer comments about the products to form a second set of consumer sentiments, the second commerce metrics including the second consumer comments, the commerce metric difference values including differences between the first set of consumer sentiments and the second set of consumer sentiments.

16. The non-transitory computer-readable storage medium of claim 14 , wherein the first commerce metrics include first ratings metrics for the products and first traffic metrics for the products, the second commerce metrics include second ratings metrics for the products and second traffic metrics for the products, and the commerce metric difference values include (a) differences between the first ratings metrics and the second ratings metrics, and (b) differences between the first traffic metrics and the second traffic metrics.

17. The system of claim 11 , wherein the first traffic metrics are based on a number of first consumer feedback instances for the products summed with a number of first consumer comments about the products, and wherein the second traffic metrics are based on a number of second consumer feedback instances for the products summed with a number of second consumer comments about the products.

18. The non-transitory computer-readable storage medium of claim 16 , wherein the first traffic metrics are based on a number of first consumer feedback instances for the products summed with a number of first consumer comments about the products, and wherein the second traffic metrics are based on a number of second consumer feedback instances for the products summed with a number of second consumer comments about the products.

19. The system of claim 1 , wherein the machine learning circuitry trains the model using a first portion of the first trend dataset, and wherein, prior to passing the second trend dataset through the model, the machine learning circuitry is further to:

run a second portion of the first trend dataset through the model;

compare an output of the model corresponding to the second portion of the first trend dataset to sales data provided to the system by the eCommerce cooperators; and

update coefficients of the model based on the comparison.

20. The system of claim 9 , further including means for determining product ratios to calculate product contribution probability indices corresponding to the products for (a) the eCommerce cooperators and (b) the eCommerce non-cooperators, the product contribution probability indices at least partially based on outputs of the model.

21. The system of claim 20 , wherein the allocating sales means is to estimate the sales missing from collected panel sales data at least partially based on the eCommerce non-cooperators and the product contribution probability indices.