Systems and methods for generating optimized market plans
Systems and methods for ranking or optimizing market configurations are disclosed. Optimized configurations of stores and formats in a market bounded within a geographic area are produced. The geographic area is divided into cells, and the cell attributes are used to determine which store formats can be present. Market configurations for each cell are provided to forecasters for determining of a first fitness criteria. Market configurations are filtered based on the first fitness criteria. Impacts between market configurations are calculated. A genetic algorithm produces a plurality of market solutions. In embodiments, optimization tasks are performed in parallel.
1. A market optimization system for determining one or more optimal market solutions for a geographic area that is divisible into a plurality of cells, each market solution assigning a market combination to each of the plurality of cells, each market combination including zero or more stores each having a format, the system comprising:
a memory;
at least one parallel processor;
a user interface including a mapping screen providing a digital representation of the geographic area;
a market definition generator configured to:
receive, via the user interface, a user defined geographic region defining a boundary of the digital representation,
subdivide the geographic region into the plurality of cells to generate a target cellular data structure,
associate at least one location attribute with each cell in the target cellular data structure, wherein each cell is representative of a subdivision of the geographic region and the at least one location attribute representative of the feasibility of the cell to be a store site,
store the target cellular data structure and the at least one location attribute associated with each cell of the target cellular data structure in the memory,
determine, cell-by-cell for the target cellular data structure using the at least one parallel processor such that at least two cells in the target cellular data structure are operated on concurrently, a set of feasible cells based on the at least one location attribute and one or more feasibility criteria,
provide a consumer population attribute, an existing store attribute and a competitor information attribute for each feasible cell, and
store the set of feasible cells in the memory as a feasible cellular data structure;
a format constraints evaluator configured to:
determine, cell-by-cell for the feasible cellular data structure using the at least one parallel processor such that at least two cells in the feasible cellular data structure are operated on concurrently, at least one possible market combination for each feasible cell in the feasible cellular data structure based on at least one of:
the at least one location attribute, the consumer population attribute, the existing store attribute, and the competitor information attribute of the feasible cell, and
at least one constraint of each format, and
store each possible market combination in the memory associated with the feasible cell;
a forecaster configured to:
determine, cell-by-cell for the feasible cellular data structure using the at least one parallel processor such that at least two cells in the feasible cellular data structure are operated on concurrently, a set of viable market combinations of each feasible cell in the feasible cellular data structure by:
predicting a first objective value for each of the at least one possible market combinations of the feasible cell, and
adding each of the at least one possible market combinations of each feasible cell to the set of viable market combinations if the first objective value of the possible market combination meets a first objective threshold, and
store the set of viable market combinations of each feasible cell in the memory associated with the feasible cell;
an impact calculator configured to:
determine an impact value for each pair of market combinations having a first market combination selected from the set of viable market combinations of a first cell and a second market combination selected from the set of viable market combinations of a second cell, and
store each impact value in a data structure in the memory associated with the first market combination and the second market combination; and
an optimizer configured to generate an output data structure including one or more legal market solutions ranked by a fitness value,
each legal market solution having a market combination assigned to each cell of the target cellular data structure whereby:
each cell that is not in the feasible cellular data structure is assigned a market combination including no stores and
each cell that is in the feasible cellular data structure is assigned a market combination that is in the set of viable market combinations for the feasible cell, and
the fitness value being based on the sum of the first objective value of each market combination in the market solution or a second objective value of each market combination, and the impact values for each pair of market combinations in the solution,
wherein generation of the output data structure comprises:
modifying a population of one or more randomly generated legal market solutions by applying one or more initial population constraints,
seeding a genetic algorithm with the population,
iterating the genetic algorithm by modifying the population based on the fitness value of each of the one or more legal market solutions of the population,
adding the legal market solutions with the highest fitness value to the output, wherein iteration of the genetic algorithm is stopped upon reaching a stopping condition, the stopping condition including a stalling threshold or a maximum number of iterations based on the one or more initial population constraints, and
updating the one or more initial population constraints based on the stopping condition reached.
2. The system of claim 1 , wherein each of the cells in the plurality of cells is a square in a grid.
3. The system of claim 1 , wherein the first objective value is one of net present value, market share, or shareholder return.
4. The system of claim 1 , wherein the optimizer comprises a plurality of parallel fitness evaluators each configured to determine a fitness value of a legal market solution, each legal market solution in the population assigned to one of the parallel fitness evaluators such that the fitness value of a plurality of legal market solutions can be determined simultaneously.
5. The system of claim 1 , wherein the forecaster comprises a plurality of parallel objective evaluators each configured to determine the first objective value for a market combination, each viable market combination assigned to one of the parallel objective evaluators such that the first objective value for a plurality of viable market combinations can be determined simultaneously.
6. The system of claim 1 , wherein the impact calculator comprises a plurality of parallel impact evaluators each configured to determine the impact value of a pair of market combinations, each pair of market combinations having a first market combination selected from the set of viable market combinations of a first cell and a second market combination selected from the set of viable market combinations of a second cell assigned to an impact evaluator such that a plurality of impact values can be determined simultaneously.
7. The system of claim 1 , wherein the one or more feasibility criteria include the availability of appropriately zoned land.
8. The system of claim 1 , wherein the first objective value for the market combination of a feasible cell is based on the consumer population attribute, the existing store information attribute and the competitor information attribute of the feasible cell.
9. The system of claim 1 , wherein the impact value is based on a projected decrease in store revenue of the first market combination caused by the presence of the second market combination.
10. The system of claim 1 , further comprising a fueling station locator configured to provide a list of fueling station locations based on a change in the first objective value predicted to result from the addition of a fueling station a selected store of the zero or more stores of a market solution, wherein a fueling station location is added to the list of fueling station locations if the change in the first objective value is positive.
11. The system of claim 1 , further comprising an online pickup locator configured to provide a list of online pickup locations based on a change in the first objective value predicted to result from the addition of an online pickup area to a selected cell of a market solution, wherein a cell location is added to the list of online pickup locations if the change in the first objective value is positive.
12. A method for determining one or more optimal market solutions for a geographic area that is divisible into a plurality of cells, each market solution assigning a market combination to each of the plurality of cells, each market combination including zero or more stores each having a format, the method comprising:
providing at least one parallel processor;
providing a user interface including a mapping screen including a digital representation of the geographic area;
receiving, via the user interface, a user defined geographic region defining a boundary of the digital representation;
subdividing the geographic region into the plurality of cells to generate a target cellular data structure;
associating at least one location attribute with each cell in the target cellular data structure, wherein each cell is representative of a subdivision of the geographic region and the at least one location attribute is representative of the feasibility of the associated cell to be a store site,
storing the target cellular data structure and the at least one location attribute associated with each cell of the target cellular data structure in the memory,
determining, cell-by-cell for the target cellular data structure using the at least one parallel processor such that at least two cells in the target cellular data structure are operated on concurrently, a set of feasible cells based on the at least one location attribute and one or more feasibility criteria as a feasible cellular data structure;
providing a consumer population attribute, an existing store attribute and a competitor information attribute for each feasible cell;
determining cell-by-cell for the feasible cellular data structure using the at least one parallel processor such that at least two cells in the feasible cellular data structure are operated on concurrently, at least one possible market combination for each feasible cell of the set of feasible cells based on at least one of:
the at least one location attribute, the consumer population attribute, the existing store attribute, and the competitor information attribute of the feasible cell, or
at least one constraint of each format;
determining cell-by-cell for the feasible cellular data structure using the at least one parallel processor such that at least two cells in the feasible cellular data structure are operated on concurrently, a set of viable market combinations of each feasible cell in the feasible cellular data structure by:
predicting a first objective value for each of the at least one possible market combinations of the feasible cell, and
adding each of the at least one possible market combinations of each feasible cell to the set of viable market combinations if the first objective value of the possible market combination meets a first objective threshold;
determining an impact value for each pair of market combinations having a first market combination selected from the set of viable market combinations of a first cell and a second market combination selected from the set of viable market combinations of a second cell, and storing each impact value in a data structure; and
generating an output data structure including one or more legal market solutions ranked by a fitness value,
each legal market solution having a market combination assigned to each cell of target cellular data structure, whereby—
each cell that is not in the feasible cellular data structure is assigned a market combination including no stores and
each cell that is in the feasible cellular data structure is assigned a market combination that is in the set of viable market combinations for the cell, and
the fitness value being based on the sum of the first objective value of each market combination in the market solution or a second objective value of each market combination, and the impact values for each pair of market combinations in the solution,
wherein generation of the output data structure is accomplished by:
modifying a population of one or more randomly generated legal market solutions by applying one or more initial population constraints,
seeding a genetic algorithm with the population,
iterating the genetic algorithm by modifying the population based on the fitness value of each of the one or more legal market solutions of the population,
adding the legal market solutions with the highest fitness value to the output, wherein iteration of the genetic algorithm is stopped upon reaching a stopping condition, the stopping condition including a stalling threshold or a maximum number of iterations based on the one or more initial population constraints, and
updating the one or more initial population constraints based on the stopping condition reached.
13. The method of claim 12 , wherein each of the one or more cells is a square in a grid.
14. The method of claim 12 , wherein the first objective value is one of net present value, market share, or shareholder return.
15. The method of claim 12 , further comprising determining the fitness value of a plurality of legal market solutions simultaneously.
16. The method of claim 12 , further comprising determining the first objective value of the plurality of possible market combinations simultaneously.
17. The method of claim 12 , further comprising determining the impact value of a plurality of pairs of market combinations simultaneously.
18. The method of claim 12 , wherein the one or more feasibility criteria include the availability of appropriately zoned land.
19. The method of claim 12 , wherein the first objective value for the market combination of a feasible cell is based on the consumer population attribute, the existing store information attribute and the competitor information attribute of the feasible cell.
20. The method of claim 12 , wherein the impact value is based on a projected decrease in store revenue of the first market combination caused by the presence of the second market combination.
21. The method of claim 12 , further comprising providing a list of fueling station locations based on a change in the first objective value predicted to result from the addition of a fueling station to a selected store of the zero or more stores of a market solution, wherein a fueling station location is added to the list of fueling station locations if the change in the first objective value is positive.
22. The method of claim 12 further comprising providing a list of online pickup locations based on a change in the first objective value predicted to result from the addition of an online pickup area to a selected cell in a market solution, wherein a cell location is added to the list of online pickup locations if the change in the first objective value is positive.