System and method for selecting promotional products for retail

1. A computer-implemented method, comprising:

receiving retail data from at least one retailer computer system;

transforming the retail data, the retail data comprising at least one selected from the group of customer data, product data, transaction data, price data, promotion data, store data and time data, the retail data corresponding to a time series having one or more time periods;

determining an at least one feature of the transformed retail data;

generating a retail promotional model for operating an intelligent agent, the retail promotional model comprising:

the at least one feature of the transformed retail data; and

a control policy for operating the intelligent agent for selecting two or more promotional products,

providing an intelligent agent module for providing the intelligent agent, the intelligent agent module comprising:

a sensor input component for receiving sensor data comprising a measured reward, wherein the sensor data is representative of a measured state of an environment, the sensor input component correcting a simulated state of the environment based on the sensor data comprising the measured reward;

a memory component comprising a long-term memory representative of a long-term frequency response, the long-term memory storing a plurality of long-term sensor data and a plurality of long-term prior actions, and a short-term memory representative of a short-term frequency response, the short-term memory storing a plurality of short-term sensor data and a plurality of short-term prior actions, the memory component storing the sensor data in the plurality of short-term sensor data of the short-term memory, the memory component storing the sensor data in the plurality of long-term sensor data of the long-term memory;

a simulation component for executing a reinforcement learning algorithm to simulate based on the short-term memory of the memory component, the long-term memory of the memory component, the simulated state of the environment, and the retail promotional model, an expected reward based on the two or more promotional products, the expected reward simulating a sales metric of a plurality of product records in the retail data based on the two or more promotional products; and

the retail promotional model.

2. The method of claim 1 , further comprising:

determining a power spectrum based on the retail data by applying a Fourier transform to the retail data; and

wherein the retail promotional model further comprises the power spectrum.

3. The method of claim 2 , wherein the determining the at least one feature of the retail data further comprises:

determining the at least one feature of the retail data for each time period based on an association rule mining function; and

wherein the at least one feature of the retail data comprises an N×N matrix.

4. The method of claim 3 , further comprising:

determining a set of relevant features from the at least one feature of the retail data based on a linear correlation of each feature in the at least one feature of the retail data with a metric.

5. The method of claim 3 , further comprising:

determining a set of relevant features from the at least one feature of the retail data based on a principal component analysis of each feature in the at least one feature of the retail data with a metric.

6. The method of claim 3 , further comprising:

determining a set of relevant features from the at least one feature of the retail data based on a least absolute shrinkage and selection operator (LASSO) analysis.

7. The method of claim 3 , further comprising:

determining a set of relevant features from the at least one feature of the retail data based on a deep neural network auto encoder analysis.

8. The method of claim 4 , wherein the set of relevant features includes one or more feature types, the one or more feature types comprising: a cannibalization type, a cadence type, an associated sales type, a price elasticity type, and a forward buying type.

9. The method of claim 8 , wherein:

the retail data further comprises a product hierarchy; and

the one or more feature types further comprises: a product hierarchy type.

10. The method of claim 9 , wherein:

the transaction data comprises point-of-sale data;

the promotion data comprises at least one selected from the group of promotion master data, promotion product master data, and promotion product customer data;

the product data comprises product master data;

the store data comprises at least one selected from the group of store data, store master data, planogram data, store grouping data.

11. The method of claim 10 , wherein the transforming the retail data further comprises:

applying a holiday adjustment function to the retail data, wherein the holiday adjustment function corrects for calendar date variations in holidays from a first time period to a second time period.

12. The method of claim 11 wherein the association rule mining function is an aprioriTID algorithm.

13. The method of claim 12 , further comprising:

determining the control policy for the retail promotional model, by,

executing a reward function based on the retail data for each time period in the one or more time periods;

selecting the at least one feature of the transformed retail data that maximize an output of the reward function for the one or more time periods; and

wherein the control policy further comprises the at least one feature of the transformed retail data.

14. A computer implemented system, comprising:

a memory, the memory having:

retail data, the retail data comprising at least one selected from the group of customer data, product data, transaction data, price data, promotion data, store data and time data, the retail data corresponding to a time series having one or more time periods; and

an intelligent agent module, the intelligent agent module for providing an intelligent agent, the intelligent agent module comprising:

a simulation component for executing a reinforcement learning algorithm to simulate based on a short-term memory of a memory component, a long-term memory of the memory component, a simulated state of an environment, and a retail promotional model, an expected reward based on two or more promotional products, the expected reward simulating a sales metric of a plurality of product records in the retail data based on the two or more promotional products;

a sensor input component for receiving sensor data comprising a measured reward, wherein the sensor data is representative of a measured state of the environment; and

the memory component comprising the long-term memory representative of a long-term frequency response, the long-term memory storing a plurality of long-term sensor data and a plurality of long-term prior actions, and the short-term memory representative of a short-term frequency response, the short-term memory storing a plurality of short-term sensor data and a plurality of short-term prior actions, and

the retail promotional model,

a processor in communication with the memory, the processor configured to:

receive the retail data from at least one retailer computer system;

transform the retail data;

determine an at least one feature of the transformed retail data;

generate the retail promotional model for operating the intelligent agent, the retail promotional model comprising:

the at least one feature of the transformed retail data; and

a control policy for operating the intelligent agent for selecting the two or more promotional products,

store the retail promotional model in the intelligent agent module;

storing the sensor data in the plurality of short-term sensor data of the short-term memory,

storing the sensor data in the plurality of long-term sensor data of the long-term memory; and

correcting the simulated state of the environment based on the sensor data comprising the measured reward.

15. The system of claim 14 wherein the processor is further configured to:

determine a power spectrum based on the retail data by applying a Fourier transform to the retail data; and

wherein the retail promotional model further comprises the power spectrum.

16. The system of claim 15 wherein the processor is further configured to:

determine the at least one feature of the retail data for each time period based on an association rule mining function; and

wherein the at least one feature of the retail data comprises an N×N matrix.

17. The system of claim 16 wherein the processor is further configured to:

determine a set of relevant features from the at least one feature of the retail data based on a linear correlation of each feature in the at least one feature of the retail data with a metric.

18. The system of claim 16 wherein the processor is further configured to:

determine a set of relevant features from the at least one feature of the retail data based on a principal component analysis of each feature in the at least one feature of the retail data with a metric.

19. The system of claim 16 wherein the processor is further configured to:

determine a set of relevant features from the at least one feature of the retail data based on a least absolute shrinkage and selection operator (LASSO) analysis.

20. The system of claim 16 wherein the processor is further configured to:

determine a set of relevant features from the at least one feature of the retail data based on a deep neural network auto encoder analysis.

21. The system of claim 17 wherein the set of relevant features includes one or more feature types, the one or more feature types comprising: a cannibalization type, a cadence type, an associated sales type, a price elasticity type, and a forward buying type.

22. The system of claim 21 , wherein:

the product data further comprises a product hierarchy; and

the one or more feature types further comprises: a product hierarchy type.

23. The system of claim 22 , wherein:

the transaction data comprises point-of-sale data;

the promotion data comprises at least one selected from the group of promotion master data, promotion product master data, and promotion product customer data;

the product data comprises product master data;

the store data comprises at least one selected from the group of store data, store master data, planogram data, store grouping data.

24. The system of claim 23 , wherein the processor is further configured to:

apply a holiday adjustment function to the retail data, wherein the holiday adjustment function corrects for calendar date variations in holidays from a first time period to a second time period.

25. The system of claim 24 wherein the association rule mining function is an aprioriTID algorithm.

26. The system of claim 24 wherein the processor is further configured to:

determine the control policy for the retail promotional model, by,

executing a reward function based on the retail data for each time period in the one or more time periods;

selecting the at least one feature of the transformed retail data that maximize an output of the reward function for the one or more time periods; and

wherein the retail promotional model further comprises the at least one feature of the transformed retail data.