Competing simulator in multi-channel retailing environment among multiple retailers
View Patent ↗A system, method and computer program product for providing the ability for retailers to devise a current channel strategy (e.g., adaptive price setting) that considers competitors in a dynamic competing environment, and that enables computing a competitive advantage of a channel. To estimate a price for selling a product j in a commerce channel comprises: a) receiving, at a processor device, real market data including sales and price history data of a product j sold by one or more retailers over one or alternate sales channels t; generating, by the processor device, a competitive advantage parameter value based on the sales and price history data; and, computing, utilizing the competitive advantage parameter value, an optimum price for a particular product to be marketed in one of the one or alternate sales channel.
1. A computer-implemented method to estimate a price for selling a product j in a commerce channel comprising the steps of:
a) receiving, at a processor device, real market data including sales and price history data of a product j sold by multiple retailers over alternate sales channels t;
b) generating, by said processor device, a competitive advantage parameter value based on said sales and price history data, said generating of the competitive advantage parameter value comprising:
receiving at input nodes of a configured neural network, a data vector including the prices and customer preferences of a product j at different times, said configured neural network representing a dynamic procedure of competition in retailer market;
propagating said data to one or more intermediate nodes of said configured neural network;
implementing a function, at said intermediate nodes, for calculating a customer preference parameter value C ij t ; and
implementing a back propagation computation of said neural network to obtain gradients for use in calculating said customer preference parameter value C ij 2
calculating the competitive advantage ψ ij k of customer segment i and product j in channel t according to:
ψ ij t (k)=max {θ 1 max{0,p j s −p j t },θ 2 max{0,C ij t −C ij s }}
where k denotes a time period, p j s (p j t ) denotes the price of product “j” in respective channel s(t), c ij t (c ij s ) denotes a calculated customer preference parameter value for respective channel t(s) of customer segment “i” while buying product “j”, and θ 1 , θ 2 are the weights of price and customer preference, respectively;
c) computing, utilizing said competitive advantage parameter value, a simulated sales volume q ij t , for a channel t in a particular price setting, said simulated sales volume used in calculating an optimum price for a particular product to be marketed in one of said alternate sales channel.
2. The method as claimed in claim 1 , wherein said simulated sales volume q ij t is computed at output nodes of said configured neural network according to:
q ij t ( k )=[adjustment factor]ψw im t
where ψ im t is said competitive advantage parameter value.
3. The method as claimed in claim 1 , further comprising: configuring said neural network by:
inputting initial parameters and sales and channel history data for training said neural network;
adaptively determining, using said sales and price history data, said customer preference parameter value for said product j, said adaptively determining including:
i) determining a difference between a simulated sales volume q ij t and a real sales volume; and, based on the difference, adjusting said one or more input parameters;
ii) producing a new simulated sales volume using said adjusted said one or more input parameters; and, iterating steps (i) and (ii) until said difference is below a threshold,
whereby said initial parameters are optimized for determining said simulated sales volume q ij t .
4. The method as claimed in claim 1 , further comprising:
performing a simulation, using said configured neural network, to dynamically optimize a price of selling a product j in sales channels in competition.
5. The method as claimed in claim 4 , wherein said dynamic price optimization of selling a product j comprises: generating an objective function which can be solved for calculating a sales volume, a profit, and a retailer market share of a t th channel.
6. The method as claimed in claim 5 , wherein said objective function is governed according to:
G
t
(
p
j
t
,
q
ij
t
)
=
∑
j
=
1
M
G
j
t
(
p
j
t
,
q
ij
t
)
=
∑
j
=
1
M
(
p
j
t
-
cos
t
j
)
∑
i
=
1
L
q
ij
t
where G t (p t j ,q ij t ) is a profit obtained by multiplying a sales quantity q ij t with a difference between price p t j and a cost cost j .
7. The method as claimed in claim 1 , further comprising:
performing a simulation , using said configured neural network, to obtain one or more resulting effects of alternate sales channels by selling product j at one or more different price settings.
8. The method as claimed in claim 7 , wherein said performing a simulation includes configuring said neural network to compute a price iteration based on channels information, sales series of different retailers, and price series of different retailers, said price iteration governed according to:
p
j
t
(
k
+
1
)
=
p
j
t
(
k
)
+
λ
t
∂
G
j
t
∂
p
j
t
(
k
-
t
tag
)
-
∑
r
∈
competitor
ξ
tr
∂
G
j
r
∂
p
j
t
(
k
-
t
tag
)
wherein r is a retailer who sells products j, p t j (k+1) is a price at point in time k, t lag denotes the time lag for competition, parameters λ t and ξ tr each describe a strategy of a retailer,
∂
G
j
r
∂
p
j
t
are gradients computed by backward propagation performed using said neural network; and, summation term
∑
r
∈
competitor
ξ
tr
∂
G
j
r
∂
p
j
t
(
k
-
t
tag
)
represents decreasing a profit of the competitors.
9. The method as claimed in claim 1 , further comprising: configuring said neural network to obtain one or more resulting effects of alternate sales channels by selling product j at one or more different price settings.
10. The method as claimed in claim 9 , further comprising using said configured said neural network to compute a price iteration governed according to:
p
j
t
(
k
+
1
)
=
p
j
t
(
k
)
+
λ
t
∂
G
j
t
∂
p
j
t
(
k
-
t
tag
)
+
λ
t
∂
G
j
s
∂
p
j
t
(
k
-
t
tag
)
-
∑
r
∈
competitor
ξ
tr
∂
G
j
r
∂
p
j
t
(
k
-
t
tag
)
wherein a price p t j (k+1) at point in time in a dynamic simulation is obtained that is optimally fit for a retailer business in competition.
11. The method as claimed in claim 1 , wherein said configured neural network is a Judge Neural Network providing a piecewise-linear model constructed by nodes having one or more min, max, sum and subtraction elements.
12. A system comprising:
a memory;
a processor unit coupled to said memory, said processor unit configured to perform a method comprising:
a) receiving, at said processor device, real market data including sales and price history data of a product j sold by multiple retailers over alternate sales channels t;
b) generating, by said processor device, a competitive advantage parameter value based on said sales and price history data, said generating of the competitive advantage parameter value comprising:
receiving at input nodes of a configured neural network, a data vector including the prices and customer preferences of a product j at different times, said configured neural network representing a dynamic procedure of competition in retailer market;
propagating said data to one or more intermediate nodes of said configured neural network;
implementing a function, at said intermediate nodes, for calculating a customer preference parameter value C ij t , said calculating said customer preference parameter value C ij t obtaining gradients by implementing a back propagation computation of said neural network;
calculating the competitive advantage of of customer segment i and product j in channel t according to:
ψ ij t ( k )=max{θ 1 max{0,p j s −p j t },θ 2max{ 0,C ij t −C ij s }}
where k denotes a time period, p j s (p j t ) denotes the price of product “j” in respective channel s(t), c ij t (c ji s ) denotes a calculated customer preference parameter value for respective channel t(s) of customer segment “i” while buying product “j”, and θ 1 , θ 2 are the weights of price and customer preference, respectively; and,
c) computing, utilizing said competitive advantage parameter value, a simulated sales volume q ij t for a channel t in a particular price setting, said simulated sales volume used in calculating an optimum price for a particular product to be marketed in one of said alternate sales channel.
13. The system as claimed in claim 12 , wherein said simulated sales volume q ij t is computed at output nodes of said configured neural network according to:
q ij t ( k )=[adjustment factor]ψ im t
where ψ im t is said competitive advantage parameter value.
14. The system as claimed in claim 12 , further comprising:
configuring said neural network by:
inputting initial parameters and sales and channel history data for training said neural network;
adaptively determining, using said sales and price history data, said customer preference parameter value for said product j, said adaptively determining including:
i) determining a difference between a simulated sales volume and a real sales volume; and, based on the difference, adjusting said one or more input parameters;
ii) producing a new simulated sales volume using said adjusted said one or more input parameters; and, iterating steps (i) and (ii) until said difference is below a threshold,
whereby said initial parameters are optimized for determining said simulated sales volume q ij t .
15. The system as claimed in claim 12 , wherein said configured neural network is a Judge Neural Network providing a piecewise-linear model constructed by nodes having one or more min, max, sum and subtraction elements.
16. A computer program product for estimating a price for selling a product j in a commerce channel, the computer program product comprising a tangible storage medium readable by a processor and storing instructions run by the processor for performing a method, the method comprising:
a) receiving, at said processor device, real market data including sales and price history data of a product j sold by multiple retailers over alternate sales channels t;
b) generating, by said processor device, a competitive advantage parameter value based on said sales and price history data, said generating of the competitive advantage parameter value comprising:
receiving at input nodes of a configured neural network, a data vector including the prices and customer preferences of a product j at different times, said configured neural network representing a dynamic procedure of competition in retailer market;
propagating said data to one or more intermediate nodes of said configured neural network;
implementing a function, at said intermediate nodes, for calculating a customer preference parameter value C ij t , said calculating said customer preference parameter value C ij t including obtaining gradients by implementing a back propagation computation of said neural network;
calculating the competitive advantage ψ ij k of customer segment i and product j in channel t according to:
ψ ij t ( k )={θ 1 max{0,p j s −p j t },θ 2 max{0,C ij t −C ij s }}
where k denotes a time period, p j s (p j t ) denotes the price of product “j” in respective channel s(t), c ij t (c ij s ) denotes a calculated customer preference parameter value for respective channel t(s) of customer segment “i” while buying product “j”, and θ 1 ,θ 2 are the weights of price and customer preference, respectively; and,
c) computing, utilizing said competitive advantage parameter value, a simulated sales volume q ij t for a channel t in a particular price setting, said simulated sales volume used in calculating an optimum price for a particular product to be marketed in one of said alternate sales channel.
17. The computer program product as claimed in claim 16 , wherein said simulated sales volume q ij t is computed at output nodes of said configured neural network according to:
q ij t ( k )=[adjustment factor]ψ im t
where ψhd im t is said competitive advantage parameter value.
18. The computer program product as claimed in claim 16 , further comprising: configuring said neural network by:
inputting initial parameters and sales and channel history data for training said neural network;
adaptively determining, using said sales and price history data, said customer preference parameter value for said product j, said adaptively determining including:
i)determining a difference between a simulated sales volume and a real sales volume; and, based on the difference, adjusting said one or more input parameters;
ii producing a new simulated sales volume using said adjusted said one or more input parameters; and, iterating steps (i) and (ii) until said difference is below a threshold,
whereby said initial parameters are optimized for determining said simulated sales volume q ij t .
19. The computer program product as claimed in claim 16 , wherein said configured neural network is a Judge Neural Network providing a piecewise-linear model constructed by nodes having one or more min, max, sum and subtraction elements.