Method and system for generating pricing recommendations

• US 8,374,906 B1
• Filed: 09/30/2008
• Issued: 02/12/2013
• Est. Priority Date: 09/30/2008
• Status: Active Grant

• Current Assignee: Zilliant, Inc.
• Original Assignee: Zilliant, Inc.
• Inventors: Williams, Loren, Zhong, Shi, Liu, Xingchu, Mookherjee, Reetabrata
• Primary Examiner(s): Hayes, John
• Assistant Examiner(s): NELSON, FREDA ANN
• Application Number:
  US12/241,782
• Time in Patent Office: 1,596 Days
• Field of Search: 705/1.1, 705/7.29, 705/7.31, 705/14.13, 705/400, 705/7.35, 705/10
• US Class Current: 705/7.35
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising

Abstract

To determine pricing recommendations for goods and service products in a business-to-business environment, a set of transaction data corresponding to a set of products are processed to generate a set of pricing recommendations optimized according to an objective. Furthermore, a set of product segments may be determined, transaction data may be associated with one or more of the product segments and a demand model and associated price elasticity may be formulated for one or more of the product segments based upon the transaction data associated with the segment. Using these formulated price elasticities, pricing recommendations for each product may be determined for each of a set of customers. Using an optimization process, price elasticities are used to determine price dependent entity goals for any combination of products, customers and sets of prices, using a mathematical objective function.

Claims

1. A computer-implemented method for computing, storing and displaying pricing recommendations optimized according to business management objectives under conditions of sparse and unreliable data, the method implemented by instructions executed by a computer, comprising the steps of:

inputting and storing in memory sales transaction data for one or more products, product price lists, product and customer segmentation specifications, business policies and business rules;

preparing the sales transaction data by assembling, organizing, segmenting products and customers according to attributes, filtering and normalizing prices to generate decision variables;

using weighted linear regression algorithms and the prepared sales transaction data, determining sets of demand models, demand model parameters and associated price elasticities corresponding to sets of products, product segments, customer and customer segments;

computing a set of reliability weights for each demand model parameter and associated price elasticity for the set of products and product segments to determine a single reliability flag for each product in the set of products for indication of price elasticity estimates deemed reliable and price elasticity estimates deemed not reliable;

combining the demand model parameters for product segments and demand model parameters for customer segments to form product/customer segment-level demand model parameters, including price elasticities that are deemed reliable and price elasticities that are deemed not reliable;

using closed form algebraic programs and mixed integer non-linear programs in the computer, optimizing price recommendation values according to designated business management objectives, business policies, business rules and price elasticities that are deemed reliable and not reliable, and determining a set of pricing recommendations for a specified set of product/customer segments that are deemed reliable and a specified set of product/customer segments that are deemed not reliable; and

wherein the step of computing a set of reliability weights further comprises calculating a set of reliability weights for each demand model based on the product of a computed R-squared regression statistic, a number of distinct decision variable values and a standard deviation of a decision value distribution, and applying hierarchical shrinkage to determine a single reliability flag for each product for indication of price elasticity estimates that are deemed reliable and not reliable.

2. A computer-implemented method for computing, storing and displaying pricing recommendations optimized according to business management objectives under conditions of sparse and unreliable data, the method implemented by instructions executed by a computer, comprising the steps of:

inputting and storing in memory sales transaction data for one or more products, product price lists, product and customer segmentation specifications, business policies and business rules;

preparing the sales transaction data by assembling, organizing, segmenting products and customers according to attributes, filtering and normalizing prices to generate decision variables;

using weighted linear regression algorithms and the prepared sales transaction data, determining sets of demand models, demand model parameters and associated price elasticities corresponding to sets of products, product segments, customer and customer segments;

computing a set of reliability weights for each demand model parameter and associated price elasticity for the set of products and product segments to determine a single reliability flag for each product in the set of products for indication of price elasticity estimates deemed reliable and price elasticity estimates deemed not reliable;

combining the demand model parameters for product segments and demand model parameters for customer segments to form product/customer segment-level demand model parameters, including price elasticities that are deemed reliable and price elasticities that are deemed not reliable;

using closed form algebraic programs and mixed integer non-linear programs in the computer, optimizing price recommendation values according to designated business management objectives, business policies, business rules and price elasticities that are deemed reliable and not reliable, and determining a set of pricing recommendations for a specified set of product/customer segments that are deemed reliable and a specified set of product/customer segments that are deemed not reliable;

wherein the step of computing a set of reliability weights further comprises calculating a set of reliability weights for each demand model based on the product of a computed R-squared regression statistic, a number of distinct decision variable values and a standard deviation of a decision value distribution, and applying hierarchical shrinkage to determine a single reliability flag for each product for indication of price elasticity estimates that are deemed reliable and not reliable; and

further comprising setting a reliable flag for each set of reliability weights that exceeds a threshold.

3. A non-transitory computer-readable storage medium with an executable instructions stored thereon, wherein the instructions controls a computer to perform a computer-implemented method for computing, storing and displaying pricing recommendations optimized according to business management objectives under conditions of sparse and unreliable data, comprising the steps of:

inputting and storing in memory sales transaction data for one or more products, product price lists, product and customer segmentation specifications, business policies and business rules;

preparing the sales transaction data by assembling, organizing, segmenting products and customers according to attributes, filtering and normalizing prices to generate decision variables;

using weighted linear regression algorithms and the prepared sales transaction data, determining sets of demand models, demand model parameters and associated price elasticities corresponding to sets of products, product segments, customer and customer segments;

computing a set of reliability weights for each demand model parameter and associated price elasticity for a set of products and product segments to determine a single reliability flag for each product in the set of products for indication of price elasticity estimates deemed reliable and price elasticity estimates deemed not reliable;

combining the demand model parameters for product segments and demand model parameters for customer segments to form product/customer segment-level demand model parameters, including price elasticities that are deemed reliable and price elasticities that are deemed not reliable;

using closed form algebraic programs and mixed integer non-linear programs in the computer, optimizing price recommendation values according to designated business management objectives, business policies, business rules and price elasticities that are deemed reliable and not reliable, and determining a set of pricing recommendations for a specified set of product/customer segments that are deemed reliable and a specified set of product/customer segments that are deemed not reliable; and

wherein the step of computing a set of reliability weights further comprises calculating a set of reliability weights for each demand model based on the product of a computed R-squared regression statistic, a number of distinct decision variable values and a standard deviation of a decision value distribution, and applying hierarchical shrinkage to determine a single reliability flag for each product for indication of price elasticity estimates that are deemed reliable and not reliable.

4. A non-transitory computer-readable storage medium with an executable instructions stored thereon, wherein the instructions controls a computer to perform a computer-implemented method for computing, storing and displaying pricing recommendations optimized according to business management objectives under conditions of sparse and unreliable data, comprising the steps of:

inputting and storing in memory sales transaction data for one or more products, product price lists, product and customer segmentation specifications, business policies and business rules;

preparing the sales transaction data by assembling, organizing, segmenting products and customers according to attributes, filtering and normalizing prices to generate decision variables;

using weighted linear regression algorithms and the prepared sales transaction data, determining sets of demand models, demand model parameters and associated price elasticities corresponding to sets of products, product segments, customer and customer segments;

computing a set of reliability weights for each demand model parameter and associated price elasticity for a set of products and product segments to determine a single reliability flag for each product in the set of products for indication of price elasticity estimates deemed reliable and price elasticity estimates deemed not reliable;

combining the demand model parameters for product segments and demand model parameters for customer segments to form product/customer segment-level demand model parameters, including price elasticities that are deemed reliable and price elasticities that are deemed not reliable;

using closed form algebraic programs and mixed integer non-linear programs in the computer, optimizing price recommendation values according to designated business management objectives, business policies, business rules and price elasticities that are deemed reliable and not reliable, and determining a set of pricing recommendations for a specified set of product/customer segments that are deemed reliable and a specified set of product/customer segments that are deemed not reliable;

wherein the step of computing a set of reliability weights further comprises calculating a set of reliability weights for each demand model based on the product of a computed R-squared regression statistic, a number of distinct decision variable values and a standard deviation of a decision value distribution, and applying hierarchical shrinkage to determine a single reliability flag for each product for indication of price elasticity estimates that are deemed reliable and not reliable; and

further comprising setting a reliable flag for each set of reliability weights that exceeds a threshold.