Ai system and method for product placement, sales forecasting and heuristic assumptions
The AI-driven system addresses the limitations of conventional product placement and sales forecasting by using a multi bin packing heuristics algorithm and computing modules to optimize product placement and formulate heuristic assumptions, resulting in improved retail operations and customer experience through enhanced data-driven strategies.
Patent Information
- Application Number
- PCT/AU2024/050526
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-05-30
- Filing Date
- 2024-05-22
- Publication Date
- 2024-12-05
AI Technical Summary
Conventional product placement algorithms in retail fail to align with individual retailer preferences, and sales forecasting methods often overlook store-specific requirements, leading to inaccurate predictions and ineffective heuristic assumptions.
An AI-driven system utilizing a multi bin packing heuristics algorithm for optimizing product placement, incorporating constraints for aesthetics, segment order, and profit/loss, along with a computing system for formulating heuristic assumptions and forecasting sales, which includes modules for data analysis and experimentation to refine strategies.
The system enhances product placement strategies, improves sales forecasting accuracy, and validates heuristic assumptions, leading to optimized retail operations and enhanced customer experience by leveraging data analytics and experimentation.
Smart Images

Figure AU2024050526_05122024_PF_FP_ABST
Abstract
Description
Al SYSTEM AND METHOD FOR PRODUCT PLACEMENT, SALESFORECASTING AND HEURISTIC ASSUMPTIONSTECHNICAL FIELD
[0001] The present invention relates to Al-driven system and methods for the retail industry. In particular, the present invention relates to an Al driven system and method for optimizing the arrangement and placement of products, forecasting sale of products and a system for facilitating the formulation of heuristic assumptions within a retail environment.BACKGROUND
[0002] The following references to and descriptions of prior proposals or products are not intended to be, and are not to be construed as, statements or admissions of common general knowledge in the art. In particular, the following prior arts discussion does not relate to what is commonly or well known by the person skilled in the art, but assists in the understanding of the inventive step of the present invention of which the identification of pertinent prior art proposals is but one part.
[0003] The retail industry encompasses businesses focused on selling goods and services directly to consumers for personal or household use. It operates across various channels, including physical stores, e-commerce platforms, and mobile apps, with a strong emphasis on meeting customer needs and delivering positive shopping experiences.
[0004] Retailers manage complex supply chains, employ marketing and merchandising strategies to drive the sales, and compete in an environment by continually innovating and adapting to changing consumer preferences.
[0005] The retail industry is driven by three main factors including product placement, sales forecasting, and adherence to heuristic(guidelines) assumptions.
[0006] Product placement refers to the strategic arrangement of items within a retail space to optimize sales and enhance the customer experience. Effective product placement involves considering factors such as visibility, accessibility, and consumer behaviour. By placing high-demand or complementary products in prominent locations, retailers can increase their chances of capturing customer attention and driving sales.
[0007] However, the placement of products or Stock Keeping Units (SKUs) within retail stores is a complex process. It involves balancing the proximity of similar or substitutable products to cater to consumer needs while maximizing shelf utilization and operational efficiency.
[0008] Retailers have used various warehousing techniques and retail strategies to attract consumers and sell inventory. One of these techniques is a planogram.
[0009] Planograms are an integral part of today's merchandising technique, as a planogram seeks to optimize product placement and maintain conscious inventory management. Planograms are widely used in the retail industry, both in small stores and large stores. Planograms are generally diagrams that map the arrangement of products within a retail space. Retailers use planograms as a tool to understand product sales and capitalize on consumers' visual focal points. Planograms allow retailers to create a focal point, understand a focal point, and therefore use a focal point to increase sales.
[0010] Placement of products on shelves can impact customer behaviour due to visibility and substitution. Placement of products on shelf will also impact sales and operational costs due to operations metrics.
[0011] The conventional product placement algorithms have either prioritized aesthetics or sales metrics, often with a huge reliance on supplier-funded shelves. However, these approaches may not fully align with the specific preferences and goals of individual retailers.
[0012] In order to optimize product placement, retailers need insights into consumer behaviour and demand. Accurate sales forecasts provide retailers with a clear understanding of which products are likely to sell well and in what quantities. By aligning product placement with sales forecasts, retailers can ensure that high-demand items are prominently displayed, maximizing the potential for sales.
[0013] Sales forecasting plays a pivotal role in the financial planning of retail businesses. It involves estimating future sales based on historical data, typically for a short-term period. However, conventional forecasting methods often overlook store-specific requirements and fail to predict sales for new products accurately.
[0014] Heuristic assumptions, such as those related to product flow, visibility, and brand adjacency, inform about product placement strategies. These assumptions are often based on insights gained from past experiences and industry best practices. By adhering to these assumptions, retailers aim to create layouts that optimize sales and enhance the overall shopping experience.
[0015] However, the effectiveness of heuristic assumptions may vary across different retail categories and store environments. Therefore, it's crucial to validate these assumptions through experimentation and data analysis to ensure that product placement strategies are effective.
[0016] Furthermore, heuristic assumptions can also influence sales forecasting models. For instance, assumptions about the impact of product visibility may be incorporated into forecasting algorithms. However, if these assumptions are inaccurate or outdated, they can lead to incorrect forecasts.
[0017] Conversely, accurate sales forecasts can provide valuable insights into the effectiveness of heuristic assumptions. By analysing sales data and comparing it to predicted outcomes, retailers can refine their heuristic assumptions and improve the accuracy of their product placement strategies.
[0018] The integration of product placement, sales forecasting, and heuristic assumptions is essential for retailers to optimize their operations and enhance the shopping experience for customers. By leveraging data analytics and experimentation, retailers can refine their strategies, adapt to changing market dynamics, and ultimately drive business success.
[0019] To address these challenges, there is a need for a system and method for product placement strategies, improving sales forecasting accuracy, and validating heuristic assumptions in a cost-effective and scalable manner.
[0020] Any discussion of the prior art throughout the specification should in no way be considered as an admission that such prior art is widely known or forms part of common general knowledge in the field.SUMMARY
[0021] PROBLEMS TO BE SOLVED
[0022] It may be an advantage to provide an Al driven integrated system for the retail environment.
[0023] It may be an advantage to provide a system and method for optimizing the arrangement and placement of products via a multi bin packing heuristics algorithm.
[0024] It may be an advantage to provide a system and method for forecasting retail sales of products.
[0025] It may be an advantage to provide a system and method for formulating heuristic assumptions for product placement, pricing strategies, marketing tactics, or customer engagement approaches.
[0026] It may be an advantage to provide a system that is capable of generating heuristics or policies which acts as an input to the core algorithms.
[0027] It may be an advantage to provide a system for forecasting sales that can assess the current availability of one or more products in each store and accordingly determine whether these stores are experiencing intermittent or regular sales of these products.
[0028] It may be an advantage to provide a system that determines which stores may be added or removed from the experiment based on one or more inputs received from a user or retailer.20-40 Meagher Street, Chippendale
[0029] It may be an advantage to provide a system, wherein the user can view results as they are received and update probability of heuristic success or failure.
[0030] It may be an advantage to provide a system that can create product placement mud map for retail stores.
[0031] It is an object of the present invention to overcome or ameliorate at least one of the disadvantages of the prior art, or to provide a useful alternative.
[0032] MEANS FOR SOLVING THE PROBLEM
[0033] The present invention is envisioned to an Al driven system and method for optimizing the arrangement and placement of products, forecasting the sale of products and facilitating the formulation of heuristic assumptions within a retail environment.
[0034] In a first aspect of the present invention, a multi bin packing heuristics algorithm is disclosed that facilitates the development of product placement mud maps for retail stores. This algorithm generates policies that serve as inputs to core algorithms. The algorithm within the system includes constraints and solutions. The constraints include unbreakable (hard) constraints, configurable hierarchical (soft) constraints, and configurable supply constraints in a binary setting. The configurable hierarchical (soft) constraints encompass aesthetic, segment order, and profit and loss constraints. Theaesthetic constraint ensures that all segments, blocks, and products are arranged in an aesthetically manner, while the segment order constraint ensures adherence to segment priority orders.
[0035] In a second aspect of the present invention, the system for formulating heuristic assumptions includes a computing system having one or more hardware processors; a memory coupled to the one or more hardware processors; a plurality of modules executable by the one or more hardware processors. The plurality of modules consists of a heuristic recommendation module to recommend a heuristic to experiment on upcoming minor range reviews; a store determination module to determine which stores may be added or removed from the experiment based on one or more inputs received from a user; an experiment performing module to automatically run the experiment using the multiarmed bandit method to minimize the costs; a data updating module to enable the user to view results as they are received and update probability of heuristic success or failure.
[0036] In a third aspect of the present invention, the system for forecasting the sale of products includes a computing system having one or more hardware processors; a memory coupled to the one or more hardware processors, further includes a plurality of subsystems executable by the one or more hardware processors; plurality of subsystems that includes a data receiver subsystem configured to receive one or more inputs; an availability determination subsystem coupled to the data receiver subsystem to determine availability of one or more products based on the received one or more inputs; a data obtaining subsystem communicatively coupled to the availability determination subsystem to obtain one or more products state based on the determined availability of the one or more products; a data generation subsystem communicatively coupled to the availability determination subsystem configured to generate one or more time series components based on the determined availability of the one or more products.
[0037] In another aspect of the present invention, a method for formulating heuristic assumptions is disclosed. The method includes recommending a heuristic to experiment on. The recommendation may span one category, or multiple similar categories. Further, the method includes the steps of ; recommending upcoming minor range reviews whichare leveraged to experiment; determining which stores may be added or removed from the experiment based on one or more inputs received from a user or retailer; automatically running the experiment using the multi-armed bandit method to minimize costs and thereafter enabling the user to view results as they are received and updating probability of heuristic success or failure.
[0038] The invention is to be interpreted with reference to the at least one of the technical problems described or affiliated with the background art of the invention. The present aims to solve or ameliorate at least one of the technical problems and this may result in one or more advantageous effects as defined by this specification and described in detail.BRIEF DESCRIPTION OF DRAWINGSThe invention will be described and explained with additional specificity and detail with the accompanying figures in which:FIG. 1 is a pictorial representation of a customer decision tree, in accordance with the embodiments in present invention;FIG. 2 is an exemplary representation of a customer decision tree from a web app view, in accordance with the embodiments in the present invention;FIG 3 illustrates an exemplary representation of the user interface in its physical form, showcasing a range of planograms based on several parameters;FIG. 4 is a block diagram illustrating a computing system for recommending heuristic assumptions, in accordance with an embodiment of the present invention;FIG. 5 is a process flow diagram illustrating an exemplary method for recommending heuristic assumptions, in accordance with an embodiment of the present invention;FIG. 6a is a block diagram illustrating a computing system for forecasting retail sales of products in accordance with an embodiment of the present invention;FIG. 6b is a block diagram illustrating various components in the computing system, in accordance with an embodiment of the present invention;FIG. 7 is a process flowchart illustrating a method for forecasting retail sales of products in accordance with an embodiment of the present invention;FIG. 8 illustrates a GANTT chart (bar chart) depicting the sales performance between April 2022 and October 2023 respectively; andFIG. 9 illustrates a block diagram that depicts the high-density and low-density runtime modes.DESCRIPTION OF THE INVENTION
[0039] For the purpose of promoting an understanding of the principles of the invention, reference will now be made to the embodiment illustrated in the figures and specific language will be used to describe them. It will nevertheless be understood that no limitation of the scope of the invention is thereby intended. Such alterations and further modifications in the illustrated online platform, and such further applications of the principles of the invention as would normally occur to those skilled in the art are to be construed as being within the scope of the present invention.
[0040] The terms "comprises”, “comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process or method that comprises a list of steps does not include only those steps but may include other steps not expressly listed or inherent to such a process or method. Similarly, one or more devices or subsystems or elements or structures or components preceded by "comprises... a" does not, without more constraints, preclude the existence of other devices, subsystems, elements, structures, components, additional devices, additional subsystems, additional elements, additional structures or additional components.
[0041] Appearances of the phrase "in an embodiment", "in another embodiment" and similar language throughout this specification may, but not necessarily do, all refer to the same embodiment.
[0042] In the following specification and the claims, reference will be made to a number of terms, which shall be defined to have the following meanings. The singular forms “a”, “an”, and “the” include plural references unless the context clearly dictates otherwise.
[0043] A computer system (standalone, client or server computer system) configured by an application may constitute a “module” that is configured and operated to perform certain operations. In one embodiment, the “module” may be implemented mechanically or electronically, so a module may comprise dedicated circuitry or logic that is permanently configured (within a special-purpose processor) to perform certain operations. In another embodiment, a “module” may also comprise programmable logic or circuitry (as encompassed within a general-purpose processor or other programmable processor) that is temporarily configured by software to perform certain operations.
[0044] Accordingly, the term “module” should be understood to encompass a tangible entity, be that an entity that is physically constructed permanently configured (hardwired) or temporarily configured (programmed) to operate in a certain manner and / or to perform certain operations described herein.
[0045] Accordingly, the term "mud map" likely refers to a visual or graphical representation of the planned product placement strategy. It could include diagrams, maps, or layouts illustrating the intended placement of products throughout the store.
[0046] The term “planogram” refers to visual representations of a store's products or services on display. They are considered a tool for visual merchandising. A planogram defines the location and quantity of products to be placed on display, often with detailed specifications on the number of product facings and spacing; shelf layout, height, width, slant and depth and necessary or recommended chiller conditions.
[0047] The term "assortment" refers to the collection or range of products available for sale within a particular category or department.
[0048] Product Placement
[0049] Embodiments of the present invention provides propose a novel approach to multi bin packing heuristics algorithm to create product placement mudmap for retail stores, which overcomes several limitations in the existing methods. The present invention introduces planogram management supporting different shelf types.
[0050] FIG. 1 is a pictorial representation 100 of a customer decision tree, in accordance with the embodiments in present invention.
[0051] A Customer Decision Tree (CDT) is a tree representation of how a consumer shops for a product in a specific category. Nodes in a CDT are called Use Cases which depict different stages or decision points in the consumer's shopping journey.
[0052] In an embodiment, a planogram is a plan for placement of products, wherein the planogram comprises of five components.
[0053] First is assortment which is a collection of products grouped according to the CDT. This ensures that products are organized in a way that aligns with how consumers make purchasing decisions.
[0054] Second is merchandising guidelines that are rules dictating the placement of products within the planogram. These guidelines determine which products are placed where to optimize visibility, accessibility, and sales.
[0055] Third is supply chain guidelines that are rules governing aspects of inventory management and replenishment, such as the number of days of inventory to keep on the shelf, delivery schedules, and order quantities.
[0056] Fourth is fixture definitions which are specifications regarding the physical layout of the retail space, including the number of sections, shelves, and their dimensions.
[0057] Fifth is placement that ensures that the assortment is placed onto sections and shelves in adherence to merchandising guidelines, maximizing the effectiveness of the planogram.
[0058] In such an embodiment, planograms are represented in the web application by uploading an assortment as an item master.
[0059] The CDT and merchandising guidelines are encoded into the following columns ({segment, block, segment priority order, block sort order, variant}). Supply chain guidelines are in the last two columns under days of supply and minimum on shelf. The assortment is determined based on expected sales / profit / dispersion, and this changes each time the sales data is updated.
[0060] FIG. 2 is an exemplary representation 200 of a customer decision tree from a web application view, in accordance with the embodiments in the present invention. The CDT is usually defined through statistical modelling. The outputs of the statistical modelling determine the best arrangement for products to increase profitability. However, a user may choose to swap layers around (for example, the merchant may state that they prefer brand / type / size over another profitable product). The user can also make commercial arrangements that “win” over the modelling (for example, a paid marketing arrangement to display one product over another in a certain position in the shelves).
[0061] Various embodiments of the present invention optimize for Profit and Loss (defined as either sales amount, profit or dispersion (as per the user selection)) by following constraints disclosed in the present invention.
[0062] Constraints comprise of unbreakable (hard) constraints, configurable hierarchical (soft) constraints and configurable supply constraints (Binary Setting).
[0063] In an embodiment, the unbreakable (hard) constraints are disclosed, wherein the products are grouped together by variant, segment and block respectively. The products that are in the same segment and block, but different variants do not have to be grouped together. The variant is considered as the top-level hierarchy for grouping products together. Segment as the second level hierarchy for grouping products together. Block is the third level hierarchy for grouping products together.
[0064] If a segment or block can fit on N shelves (if N>1), it can’t span more than N. eg if a segment / block can fit on 2 shelves, it can’t fill 3 shelves. In a case where N=l, Segments may be allowed to span 2 shelves even if they fit on 1 shelf depending on the configurable constraints. A segment or block can only span M Fixtures if at least (M-l) of the fixtures is 80% allocated to the segment. Two segments that have the same segment priority order should always be placed beside right after another. If segment A and segment B both have segment priority order=1.0, which one to place first? Whichever one is feasible given the configurable constraints. If both are feasible, then whichever one generates the highest profit and loss will be given the priority. Within a block, products must be sorted by block sort order. Classification of adjacency starts counting from the left on odd number shelves, starts on the right for even number shelves.
[0065] Table 1 is an illustration of 14 products laid out on 2 shelves that should be thought of as being sequenced. In the table product 7 and product 8 are adjacent, but products 7 and 14 are not adjacent. Products fill all shelves and none of the shelves are empty. The same product ID must be continuously displayed and can’t span sections. The same product ID can’t appear in more than one section.
[0066] In an exemplary embodiment, products within a segment or block must be continuous and form either a rectangle or rectangle + half a shelf (with a small tolerance). Tolerance should be in % and should be configurable. It is always OK to span exactly N shelves or N.5 shelves, where N is an integer if a segment / block is only on 1 section (See Examples 1, 2 and 3 below). It is always OK to span exactly 0.5+0.5 shelves if a segment / block is only on 1 section (See Example 4 below).Table 2
[0067] It is not OK to have 0.5+0.5 shelves if a segment or block spans 2 sections. (See Example 5 below). It is not OK to end up in either of the strange “Tetris” shapes. (See Example 6a and 6b below). In example 6a segment / block is assigned 1.7 shelves. The segment or block could fit on 2 shelves but spans 3. Example 6b violates 2 conditions. Segment / Block is assigned 2.7 shelves. First, Products are not placed as a rectangle. Second, the Segment could fit on 3 Shelves but spans 4 Shelves instead.Table 3
[0068] It is not OK to Span N sections if N-l sections are not at least 80% full, and if the last section does not contain at least 1 full shelf. For example, it is not OK to span 3 sections if 2 sections are not at least 80% full with 1 full shelf in 3rd shelf. (See Example 7 below). If a segment or block is an awkward shape, they can be rearranged or trimmed until the shape is acceptable wherein rearranging segments that belongs to different segment priority order is not possible if segment order constraint is higher than profit and loss (P&L) constraint. Trim should minimise lost P&L. When given a choice to trim segment A or segment B, the present invention chooses to trim the one that loses least amount of P&L.> ._ __Exampie 7 NOT OKTable 4
[0069] In another embodiment, the configurable hierarchical (soft) constraints are disclosed. The configurable hierarchical (soft) constraints comprise of aesthetic constraint, segment order constraint and profit and loss (P&L) constraint. Aesthetics constraint ensures all segments, blocks and products are arranged in an aesthetically pleasing way.
[0070] For example, if a segment can fit on A section, it can’t span A+l sections. If a block can fit on B shelves, it can’t span B+l shelves. If a product can fit on C shelf, it can’t span C+l shelf. Segment order constraint ensures all segment priority orders are adhered. Products with segment priority order 3.1 should come before priority order 3.3.
[0071] P&L constraint ensures if P&L constraint is selected as most important out of configurable constraints, then the user is allowed to break days of supply (DOS), minimum on shelf (MOS), and Aesthetic constraints to optimize P&L.
[0072] The user can specify:TOP Constraint: AESTHETICS2nd Constraint: P&L3rd Constraint: SEGMENT ORDERORTOP Constraint: P&L2nd Constraint: AESTHETICS3rd Constraint: SEGMENT ORDER
[0073] Regardless of configurable constraint hierarchy, the present invention always tries and satisfies all constraints where possible. The hierarchy is purely to allow conflict resolution when multiple constraints can’t be satisfied at the same time.
[0074] In an exemplary embodiment having P&L as top constraint. P&L is most important. The present invention does not place segments+blocks by N.5 shelves
[0075] If a segment or block gets assigned, a portion of a shelf can be assigned to a segment if that segment gets > 1 whole shelf (example 4a). So, if 1.4 shelves are allowed. 2.3 shelves are also allowed.
[0076] A fractional portion of 2 shelves can be assigned to a segment if that segment is assigned < 1 whole shelf of space (example 4b). However, the difference in width of segment assigned to each shelf should be no more than 20cm.
[0077] Example 8a-8c are OK in this configuration, while Example 9 is not OK because the difference in width of segment assigned to each partial shelf is > 20cm.
[0078] Example 10 is also not allowed because not N-l fixtures have >80% of space allocated to the segment.
[0079] Example 11 is allowed because N-l fixtures have at least 80% of space allocated to the segment, and there is at least 1 full shelf on the last section.Table 5
[0080] Example 12 is allowed because N-l sections have at least 80% of space allocated to the segment and there is at least 1 full shelf on the last section.Table 6
[0081] If a segment or block is an illegal shape, the present invention tries rearranging the segment ordering. If rearranging segments and blocks can’t satisfy the constraints, then it trims the products to minimize P&L. Segment priority order does not matter so long as segments with identical segment priority order still remain together.
[0082] Example 9 segment gets assigned < 1 shelf but is split across 2 shelves, and there is a difference in assigned segment width between the 2 shelves. This is only allowable if the aesthetics constraint is lower than the P&L constraint and Segment Order constraint. The user needs this to allow segment order to remain in priority. If a segment or block is an illegal shape, try rearranging the segment ordering. If rearranging segments and blocks can’t satisfy the constraints, then trim products to minimise. Segment priority order does not matter so long as segments with identical segment priority order still remain together.
[0083] In another exemplary embodiment having aesthetics as top constraint, wherein the user wants the layout to look aesthetically pleasing. All segments, blocks must be perfectly rectangle or rectangle + 0.5 shelves.
[0084] Examples, 4a, 4b, and 4c (above) are not allowed if aesthetics constraint is higher than P&L. If aesthetics is the top hierarchy constraint, any segment that can fit on 1 section cannot span 2 sections. Any segment that can fit on 2 sections also cannot spanacross 3 sections. In addition, all other hard constraints still apply. It is not OK to span 2 sections if the segment can fit entirely on 1 section (See Example 13).Table 7
[0085] In a case, when can a segment span more than 1 fixture (See Example 14a and 14b). Section 1 has to be 100% filled with the segment. Section 2 has to have at least 2 full shelves assigned to the segment.
[0086] If a segment or block is an illegal shape, the present invention rearranges the segment ordering. If rearranging segments and blocks can’t satisfy the constraints, then trim products to minimise P&L. Segment priority order does not matter so long as segments with IDENTICAL segment priority order still remain together.Table 8
[0087] In a further exemplary embodiment where segment priority constraint is the highest constraint, Segments must be arranged in a way such that the products with the lowest number in segment priority order appear first. In a segment / block shape hard constraints still apply and hence it is not OK to end up in a strange tetris shape (See Example 15a and 15b). Products are trimmed such that the segment and blocks meet minimum hard constraints + the supply constraint (whichever is chosen). If a segment or block is an illegal shape, the present invention does rearrange segment ordering however it trim products from segments until all constraints are met.
[0088] In an additional embodiment when a variant is present, the present invention minimizes the distance between segment A with no variant, and Segment A with a variant. Distance is defined as shelves in between the segment As. When a new variant begins, products should be placed from left to right not snake from previous shelves.
[0089] In a further embodiment, configurable supply constraint: binary setting is disclosed. Supply configurable constraints comprises of DOS constraint and MOS constraint.
[0090] DOS constraint ensures if a product is selected, the product ID must have enough facings to sell for N days (N=days_of_supply) wherein Facings= quantity sold per week / 7* days_of_supply / ( depth of shelf / depth of product). If a product cannot fit on shelf with at least DOS, it will not be ranged.
[0091] MOS constraint ensures if a product is selected, the product ID must have M units on shelf Facings= MOS / (depth of shelf / depth of product). If a product can’t fit on shelf with at least MOS, it will not be ranged. If both the constraints are selected, each product must have the max of (DOS and MOS). If a product cannot fit on shelf with at least max (DOS, MOS), it will not be ranged. If neither are selected, the present inventionmakes best attempts to place at least max (DOS and MOS) on shelf. Products can be placed on shelf with at least _SUPPLY_TOLERANCE_% of max(DOS, MOS). If DOS is selected, then all products selected will have at least DOS days minimum. Perhaps MOS is sometimes violated if MOS< DOS for any product. Its vice versa if MOS constraint is selected. Both will result in the smallest number of unique products selected with the lowest P%L. If neither option is selected, then we should have a greatest variety of products and higher P&L.
[0092] FIG 3 illustrates an exemplary representation of the user interface (dashboard) in its physical form, displaying a range of real time planograms in different colours based on various parameters / metrics such as P&L, Retailer Merchant Guideline, and Visual Impact.
[0093] This Al-powered system serves as a dynamic retail planogram optimization tool, accessible to users via tablets, phones, or computers serving as hardware-like tools.
[0094] The system operates through an interaction loop, wherein users input specific instructions or preferences, Al processes using advanced algorithms to analyze and interpret their requirements. Subsequently, the system generates results or recommendations on the user interface / platform.
[0095] Users can input various parameters, including defining shelving and merchandising guidelines, customizing decision tree parameters to override default settings, and setting limits on personnel hours allocated to planogram development tasks.
[0096] Once inputs are provided, the system dynamically generates store-level assortments by analysing customer behaviour, geographical factors, and proximity to competitors. It also generates customer decision trees to guide assortment selection based on predefined criteria.
[0097] Further, the system autonomously creates store-level planograms to key performance indicators like profitability and visual impact, accessible via tablet or relatedcomputer interfaces. The system predicts and scores planogram outcomes based on predefined metrics, facilitating data-driven decision-making.
[0098] Various embodiments of the present invention can be applied to retail sales management and planogram construction. A planogram is a schematic tool retailers use to plan their store layout to maximize sales and customer experience. Planograms place special attention on product placement and displays as well as point-of-sale (POS) location(s).
[0099] The present invention supports different shelf types (imagine a fridge vs a regular shelf vs a hangcell) and small breakage of days of supply and minimum on shelf constraints to potentially increase P&L.
[0100] Heuristic Assumptions
[0101] FIG. 4 is a block diagram illustrating a computing system 100 for recommending heuristic assumptions. The computing system 100 comprises one or more hardware processors 102, a memory 104 and a storage unit 106. The one or more hardware processors 102, the memory 104 and the storage unit 106 are communicatively coupled through a system bus 108 or any similar mechanism. The memory 104 comprises a plurality of modules 110 in the form of programmable instructions executable by the one or more hardware processors 102.
[0102] Further, the plurality of modules 110 includes a heuristic recommendation module 112, a review recommendation module 114, a store determination module 116, an experiment performing module 118 and a data updating module 120.
[0103] The heuristic recommendation module 112 is configured to recommend a heuristic to experiment on. It may utilize machine learning algorithms, statistical analysis, or predefined rules to identify promising heuristics that could potentially improve the outcomes of the range reviews. The recommendations generated by this module serve as the basis for the experimentation conducted by the system.
[0104] The recommendation may span one category, or multiple similar categories. A heuristic assumption is recommended that has the lowest cost of testing. Further, a heuristic assumption is recommended that has a highest probability of being wrong. The heuristic assumption is recommended that has some probability of being wrong, but may have high impact in the profit and loss for the retailer if they are experimented.
[0105] The review recommendation module 114 is configured to recommend upcoming minor range reviews which are leveraged to the experiment.
[0106] The store determination module 116 is configured to determine which stores may be added or removed from the experiment based on one or more inputs received from a user or retailer. Criteria for store selection may include factors such as store size, location, demographic profile of customers, historical sales performance, or any other relevant parameters. By determining the subset of stores to include in the experiment, this module helps ensure that the experiment is conducted on a representative sample of the overall store network.
[0107] The experiment performing module 118 is configured to automatically run the experiment using the multi-armed bandit method to minimize costs. In an embodiment of the present invention, the multi-arm bandit method is used for statistical testing rather than A / B testing to minimize costs. The multi-armed bandit method is a type of reinforcement learning algorithm that dynamically allocates resources (in this case, likely product placements or range adjustments) to different options (heuristics).
[0108] The data updating module 120 is configured to enable the user to view results as they are received and update probability of heuristic success or failure. Users can view metrics such as sales performance, customer feedback, or other relevant indicators to assess the effectiveness of each heuristic being tested.
[0109] By facilitating iterative refinement of the heuristics being tested, this module supports adaptive decision-making and continuous improvement throughout the range review process.
[0110] The one or more hardware processors 102, as used herein, means any type of computational circuit, such as, but not limited to, a microprocessor unit, microcontroller, complex instruction set computing microprocessor unit, reduced instruction set computing microprocessor unit, very long instruction word microprocessor unit, explicitly parallel instruction computing microprocessor unit, graphics processing unit, digital signal processing unit, or any other type of processing circuit.
[0111] The memory 104 may be non-transitory volatile memory and non-volatile memory. The memory 104 may be coupled for communication with the one or more hardware processors 102, such as being a computer-readable storage medium. The one or more hardware processors 102 may execute machine-readable instructions and / or source code stored in the memory 104. A variety of machine -readable instructions may be stored in and accessed from the memory 104. The memory 104 may include any suitable elements for storing data and machine-readable instructions, such as read only memory, random access memory, erasable programmable read only memory, electrically erasable programmable read only memory, a hard drive, a removable media drive for handling compact disks, digital video disks, diskettes, magnetic tape cartridges and memory cards. In the present embodiment, the memory 104 includes the plurality of modules 110 stored in the form of machine -readable instructions on any of the above-mentioned storage media and may be in communication with and executed by the one or more hardware processors 102.
[0112] The storage unit 106 may be a cloud storage or a local file directory within a remote server. The storage unit 106 may store upcoming minor range reviews, results, probability of heuristic success or failure and the like.
[0113] FIG. 5 is a process flow diagram illustrating an exemplary method for recommending heuristic assumptions, in accordance with an embodiment of the present invention.
[0114] At step 202, a heuristic is recommended to experiment on. In an embodiment of the present invention, the recommendation may span one category, or multiple similar categories. In an embodiment of the present invention, a category is defined as a subset of similar products in retail, such as baking, pasta, fresh produce and the like.
[0115] In an embodiment of the present invention, a heuristic assumption is recommended that has the lowest cost of testing. Further, a heuristic assumption is recommended that has the highest probability of being wrong.
[0116] The heuristic assumption is recommended that has some probability of being wrong, but may have high impact in profit and loss (P&L) for the retailer if they are experimented.
[0117] At step 204, upcoming minor range reviews are recommended which are leveraged to the experiment.
[0118] At step 206, it is determined that which stores may be added or removed from the experiment based on one or more inputs received from a user or retailer.
[0119] At step 208, the experiment is automatically run by using the multi-armed bandit method to minimize costs.
[0120] Further, at step 210, the user is enabled to view results as they are received and the probability of heuristic success or failure is updated.
[0121] The method 200 may be implemented in any suitable hardware, software, firmware, or combination thereof.
[0122] Sale Forecasting system
[0123] Embodiments of the system for forecasting the sale of products discloses a computing system 10, primarily a data forecasting system that not only forecasts retail sales of one or more products, but also use these forecasts and substitution models to determine what products to place in each of one or more stores. The computing system 10 uses machine learning methods to forecast sales of one or more products.
[0124] In one embodiment, a data validation process is performed. In the data validation process, the computing system 10 is set to initiate a quick run of the system from end to end for checking whether the input data generate results. It also checks whether it is feasible to use the computing system 10 on given assumptions entered by a user 70.
[0125] FIG. 6a is a block diagram illustrating a computing system 10 for forecasting retail sales of one or more products in accordance with an embodiment of the present invention. The computing system 10 forecasts retail sales of one or more products at a store level to determine what products to place on each of one or more stores.
[0126] FIG. 6b is a block diagram illustrating various components in the computing system.
[0127] The computing system 10 includes a hardware processor 100. The computing system 10 also includes a memory 80 coupled to the hardware processor 100. The memory 80 comprises a set of program instructions in the form of a plurality of subsystems and configured to be executed by the hardware processor 100.
[0128] The plurality of subsystems includes a data receiver subsystem 20. The data receiver subsystem 20 is configured to receive one or more inputs associated with uniqueness of the one or more products. The inputs received by this subsystem includes data such as historical sales data, inventory levels, pricing information, market trends, and any other relevant variables that could impact product sales.
[0129] In an exemplary embodiment, user 70 is required to forecast sales of a product X. The product X is fed to the computing system 10. The computing system 10 determines whether the product is new or is / was already available in market. The unique information regarding the product X is returned by the data receiver subsystem 20.
[0130] The plurality of subsystems also includes an availability determination subsystem 30. The availability determination subsystem 30 is communicatively coupled to the data receiver subsystem 20. The availability determination subsystem 30 is configured to determine availability of the one or more products in market based on the received one or more inputs.
[0131] In another exemplary embodiment, the availability of the product X is determined at the store level. It is determined that whether the one or more products are currently available at each of the one or more stores or the one or more products were sold previously at each of the one or more stores.
[0132] The plurality of subsystems further includes a data obtaining subsystem 40. The data obtaining subsystem 40 is communicatively coupled to the availability determination subsystem 30. The data obtaining subsystem 40 is configured to obtain one or more products state based on the determined availability of the one or more products.
[0133] In one embodiment, one or more equations are generated to determine the current state of the one or more products. The one or more equations comprises of the questions including whether the one or more products was on promotion recently, whether it is a consistent seller, whether the one or more products was out of stock recently, visible area, low visibility area, sells regularly (sells at least Y units per week, does not sell at least Y units per week) and the like. The response to the one or more equations determines the one or more products state. The subsystem ensures that the forecasting model incorporates the most recent and accurate data available, minimizing errors and enhancing forecast reliability.
[0134] The plurality of subsystems includes a data generation subsystem 50. The data generation subsystem 50 is communicatively coupled to the availability determination subsystem 30 and the data obtaining subsystem 40. The data generation subsystem 50 is configured to generate one or more time series components based on the determined availability of the one or more products. Further, the data generation subsystem 50 is configured to generate one or more product attributes upon generating the one or more time series components.
[0135] The data generation system employs statistical techniques, machine learning algorithms, or mathematical models to analyse historical trends, identify patterns, and predict future sales performance. It may be able to generates forecasts for individual products or product categories, taking into account factors such as seasonality, demand variability, market trends, and promotional effects.
[0136] In an embodiment, the computing system 10 incorporates elements of time series forecasting. The elements of time series forecasting comprises of geography (population density), distance to competitors, census data and the like. The computing system 10 calculates similar products sales indices in recent grid and adjusts it with time since last sale of the one or more products.
[0137] In another embodiment, the computing system 10 incorporates one or more attributes to enrich plurality of generated data by time series components and the plurality of one or more products state.
[0138] The one or more attributes comprises product attributes generated by Natural Language Processing (NLP) tokenization, store attributes generated by geographical vectors, competition vectors, weather-based attributes, customer personas and the like.
[0139] The plurality of subsystems also includes a data forecasting subsystem 55. The data forecasting subsystem 55 is communicatively coupled to data obtaining subsystem 40 and the data generation subsystem 50.
[0140] The data forecasting subsystem 55 is configured to forecast retail sales of the one or more products based on the generated one or more time series components, the one or more product attributes and the one or more products state by using data forecasting methods based on Artificial Intelligence models.
[0141] In one embodiment, the plurality of generated data is utilized to forecast sales of the one or more products through machine learning models including model tuning, input trimming, gradient descent and gradient boosting wherein gradient boosting is a popular boosting algorithm. In gradient boosting, each predictor corrects its predecessor’s error. The weights of the training instances are not tweaked, instead, each predictor is trained using the residual errors of predecessor as labels. Gradient Descent is an optimization algorithm to train machine learning models by means of minimizing errors between actual and expected results.
[0142] In a further embodiment, several machine learning modules comprising Croston and EWMA are applied to the obtained one or more products state. Croston is applied to the obtained one or more products state having intermittent sales whereas EWMA is applied to the obtained one or more products state having regular sales. Croston's method involves using simple exponential smoothing (SES) on the non-zero elements of the time series and a separate application of SES to the times between non-zero elements of the time series. The smoothing parameters of the two applications of SES are assumed to be equal and are denoted by alpha. EWMA applies weights to the values of a time series. More weight is applied to more recent data points, making them more relevant for future forecasts.
[0143] In such embodiments, all forecasts are made without space / merchandising considerations. Further, all forecasts made without final optimization parameters in mind. Cannibalisation curves are often assumed if not enough experimentation from retailers. Impact of visibility index on top products, niche products, etc. are assumed. Accurate forecasts do not imply optimization on labour.
[0144] The plurality of subsystems also includes a product determination subsystem 60. The product determination subsystem 60 is communicatively coupled to the data forecasting subsystem 55.
[0145] The product determination subsystem 60 determines the one or more products required to be placed on each of the one or more stores based on the forecast retail sales of the one or more products and one or more substitution models.
[0146] FIG. 7 is a process flowchart 120 illustrating a method for forecasting retail sales of products in accordance with an embodiment of the present invention.
[0147] At step 130, the method comprises receiving data of the one or more products and determining whether the one or more products is new in the market, performed by the data receiving subsystem 20. Upon receiving the output, if the one or more products is new in the market, it proceeds to step 190 and if the one or more products is / was present in the market, it proceeds to step 140.
[0148] At step 140, the method comprises determining the current availability of the one or more products at each of the one or more stores by the availability determination subsystem 30. If the one or more products has been previously sold at the one or more stores, it proceeds to the step 190 (generates time series components); and if the one or more are currently available for sale at the one or more stores, it proceeds to step 150.
[0149] At step 150, the one or more equations are generated to determine the current state of the one or more products by the data obtaining subsystem 40. The one or more equations comprises of the questions including was the one or more products on promotion recently, whether it is a consistent seller, and the like. Upon receiving the output of the one or more equations, it proceeds to step 160.
[0150] At step 160, the method comprises determining whether the one or more stores are producing intermittent sales or regular sales of the one or more products. In case ofintermittent sales, it proceeds to step 170 and in case of regular sales, it proceeds to step 180.
[0151] At step 170, the method comprises applying Croston forecasting model for forecasting the retail sales of the one or more products by the data forecasting subsystem 55.
[0152] At step 180, the method comprises applying EWMA forecasting model for forecasting the retail sales of the one or more products by the data forecasting subsystem 55.
[0153] At step 190, the method comprises generating time series components by the data generation subsystem 50.
[0154] At step 200, the method comprises enhancing the output of step 190 by enriching it with other attributes comprising product attributes, customer personas, and the like by the data generation subsystem 50.
[0155] At step 210, the method comprises forecasting the retail sales of the one or more products at each of the one or more stores by the data forecasting subsystem 55 upon receiving the output from the step 170, 180 and 200.
[0156] At step 220, the method comprises determining the adjustments required to availability of the one or more products at each of the one or more stores in accordance with the forecasted retail sales of the one or more products by the product determination subsystem 60.
[0157] EXAMPLES
[0158] Case Study 1 - Forecasting with fuzzy assumptions
[0159] The study focuses on forecasting with fuzzy assumptions to develop predictive models tailored to the specific needs of retailers. It provides retailers with accurate forecasts to optimize business strategies and improve decision-making processes.
[0160] Methodology
[0161] Retailer- specific data is collected through scraping techniques and processed using NLP to extract relevant information.
[0162] Data is categorized into numerical bins based on contextual factors such as product size, price, and category.
[0163] Automated processes generate inputs for forecasting models, ensuring efficiency and scalability.
[0164] Advanced forecasting techniques are applied to predict future trends and demand patterns.
[0165] Fuzzy logic controllers are utilized to accommodate uncertain or imprecise inputs, enhancing the adaptability of the models.
[0166] The methodology involves an iterative approach, refining models based on feedback and insights gathered during the forecasting process.
[0167] This study results in more accurate forecasts tailored to the specific needs of retailers. By accurately predicting demand trends and optimizing business strategies, the study adds commercial value for retailers, leading to increased profitability and competitiveness in the market.
[0168] In Figure 8, a Gantt chart (bar chart) visualizes the sales performance between April 2022 and October 2023. The chart illustrates the sales trend over an 8-week rollingperiod, The present system's performance is represented by the green bars, while the control stores' performance is depicted by the orange bars.
[0169] According to the chart, the present system consistently outperforms the control stores throughout the depicted period. Notably, in October 2023, the present system achieved approximately 13% more sales compared to the control stores. This indicates a significant improvement in sales performance attributed to the present system's implementation.
[0170] Case Study 2 - Bin packing optimization using heuristics
[0171] This study aims to develop efficient algorithms to optimize the arrangement of products on shelves in retail stores and the impact of merchandising changes on labor costs. These heuristics include reserved shelves for signage, laws governing the placement of promotions, and specific treatment of private label products.
[0172] Methodology:
[0173] Retailer-specific data, including shelf dimensions, product sizes, and merchandising rules, is collected and analyzed.
[0174] Heuristic -based algorithms are developed to optimize bin packing, considering factors such as product size, shelf space constraints, and merchandising rules.
[0175] The algorithms are refined to allow for finer control over bin packing outcomes, addressing retailer- specific requirements.
[0176] The study implements high-density and low-density runtime modes to accommodate heavily promoted products and regular business-as-usual aisles. Figure 9 is a block diagram that depicts the high-density and low-density runtime modes.
[0177] The study incorporates user feedback into the algorithm development process, allowing users to provide loose preferences and refine the bin packing outcomes iteratively.
[0178] Outcome: The study results in improved bin packing algorithms tailored to the specific requirements of retailers like Woolworths to enhance operational efficiency and user satisfaction in retail store merchandising.
[0179] Case Study 3 - Optimization of Shelf Space Allocation
[0180] The study focuses on optimizing shelf space allocation in retail stores, particularly in the Diet and Sport category with Woolworths, using advanced data-driven methodologies and considering various factors such as shelf dimensions, product dimensions, and customer preferences.
[0181] Methodology:
[0182] A data review framework is developed to analyze and track progress in the Diet and Sport category with Woolworths. This framework helps in understanding the shelf space requirements and dimensions of products.
[0183] The shelf space allocation is automated for the Diet and Sport category, taking into account various dimensions of shelves and products that involves developing algorithms based on factors such as product demand and profitability.
[0184] Furthermore, the planograms are optimised considering factors like return on space, shelf alignment, core range, and operational metrics.
[0185] The expansion to other retailers involves increasing the number of products included in planograms and leveraging learnings from existing commercial relationships to optimize shelf space allocation.
[0186] The study results in more efficient utilization of shelf space in retail stores, leading to increased visibility and accessibility of products in the Diet and Sport category. Automation of shelf space allocation processes streamlines category management activities, reducing manual effort and improving accuracy in planogram optimization.
[0187] The embodiments herein can comprise hardware and software elements. The embodiments that are implemented in software include but are not limited to, firmware, resident software, microcode, etc. The functions performed by various modules described herein may be implemented in other modules or combinations of other modules. For the purposes of this description, a computer-usable or computer readable medium can be any apparatus that can comprise, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.
[0188] The medium can be an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system (or apparatus or device) or a propagation medium. Examples of a computer-readable medium include a semiconductor or solid-state memory, magnetic tape, a removable computer diskette, a random-access memory (RAM), a read-only memory (ROM), a rigid magnetic disk and an optical disk. Current examples of optical disks include compact disk-read only memory (CD-ROM), compact disk-read / write (CD- R / W) and DVD.
[0189] Input / output (I / O) devices (including but not limited to keyboards, displays, pointing devices, etc.) can be coupled to the system either directly or through intervening I / O controllers. Network adapters may also be coupled to the system to enable the data processing system to become coupled to other data processing systems or remote printers or storage devices through intervening private or public networks. Modems, cable modem and Ethernet cards are just a few of the currently available types of network adapters.
[0190] A representative hardware environment for practicing the embodiments may include a hardware configuration of an information handling / computer system in accordance with the embodiments herein. The system herein comprises at least one processor or central processing unit (CPU). The CPUs are interconnected via system bus108 to various devices such as a random-access memory (RAM), read-only memory (ROM), and an input / output (I / O) adapter. The I / O adapter can connect to peripheral devices, such as disk units and tape drives, or other program storage devices that are readable by the system. The system can read the inventive instructions on the program storage devices and follow these instructions to execute the methodology of the embodiments herein.
[0191] The system further includes a user interface adapter that connects a keyboard, mouse, speaker, microphone, and / or other user interface devices such as a touch screen device (not shown) to the bus to gather user input. Additionally, a communication adapter connects the bus to a data processing network, and a display adapter connects the bus to a display device which may be embodied as an output device such as a monitor, printer, or transmitter, for example.
[0192] A description of an embodiment with several components in communication with each other does not imply that all such components are required. On the contrary, a variety of optional components are described to illustrate the wide variety of possible embodiments of the invention. When a single device or article is described herein, it will be apparent that more than one device / article (whether or not they cooperate) may be used in place of a single device / article. Similarly, where more than one device or article is described herein (whether or not they cooperate), it will be apparent that a single device / article may be used in place of the more than one device or article, or a different number of devices / articles may be used instead of the shown number of devices or programs. The functionality and / or the features of a device may be alternatively embodied by one or more other devices which are not explicitly described as having such functionality / features. Thus, other embodiments of the invention need not include the device itself.
[0193] Although the invention has been described with reference to specific examples, it will be appreciated by those skilled in the art that the invention may be embodied in many other forms, in keeping with the broad principles and the spirit of the invention described herein.
[0194] The present invention and the described preferred embodiments specifically include at least one feature that is industrial applicable.
Claims
Claims:
1. A system adapted for advanced planogram management and optimization, comprising; a. a user interface displaying a plurality of planograms, wherein the planograms consist of; b. a multi bin packing heuristics algorithm tailored to create product placement mud maps for retail stores; c. an assortment management component, wherein the component groups the products according to a Customer Decision Tree (CDT) to align with consumer shopping behaviours; d. a merchandising rule mechanism, wherein the mechanism for product placement optimizes customer engagement and drive sales; e. a supply chain component for addressing critical inventory aspects including but not limited to shelf inventory duration, delivery schedules, and unit quantities to optimize supply chain efficiency; and f. plurality of algorithms to intelligently place assortment items on shelves while adhering to merchandising guidelines and optimizing product visibility and accessibility.
2. The system of Claim 1, wherein the algorithm comprises of constraints and solutions.
3. The system of Claim 2, wherein the constraints include unbreakable (hard) constraints, configurable hierarchical (soft) constraints, and configurable supply constraints in a binary setting.
4. The system of Claim 3, wherein the configurable hierarchical (soft) constraints encompass aesthetic, segment order, and profit and loss constraints.
5. The system of Claim 3 or 4, wherein the aesthetic constraint ensures that all segments, blocks, and products are arranged in an aesthetically manner and segment order constraint ensures adherence to segment priority orders.
6. The system of Claim 3, wherein the unbreakable (hard) constraints groups the together by variant, segment and block respectively.
7. The system of Claim 6, wherein the variant is the first level hierarchy, segment is the second level hierarchy and block is the third level hierarchy for grouping products together.
8. The system of Claim 1, wherein the user interface includes customizable dashboards and reporting tools to track key performance indicators, monitor planogram compliance, and measures return on investment.
9. The system of Claim 3, wherein the supply configurable constraints comprise of DOS constraint and MOS constraint.
10. A system adapted for formulating heuristic real-world assumptions within a retail environment, comprising; a. one or more hardware processors facilitating the computational operations of the system; b. a memory coupled to the one or more hardware processors, wherein the processors serve as repository of data and includes a plurality of modules, consisting of; c. a heuristic recommendation module configured to analyse data and recommend a heuristic to experiment on; d. a review recommendation module that recommends upcoming minor range reviews which are leveraged to the experiment; e. a store determination module equipped with sophisticated decision-making capabilities to determine which stores may be added or removed from the experiment based on one or more inputs received from a user; f. an experiment performing module configured to automatically execute the experiment; g. a data updating module designed to provide users with real-time access to view experimental results and update probability of heuristic success or failure accordingly.
11. The system of Claim 10, wherein one or more hardware processors is a computational circuit including but not limited to a microprocessor unit, microcontroller, complex instruction set computing microprocessor unit, reduced instruction set computing microprocessor unit, graphics processing unit, digital signal processing unit, or any other type of processing circuit.
12. The system of Claim 10, wherein the memory may be non-transitory volatile memory and non-volatile memory.
13. The system of Claim 10, wherein the system includes a storage unit which may be a cloud storage or a local file directory within a remote server.
14. The system of Claim 10, further includes a method for recommending heuristic assumptions having the following steps; a. recommending a heuristic to experiment on and upcoming minor range reviews leveraged to the experiment; b. determining which stores may be added or removed from the experiment based on one or more inputs received from a user or retailer; c. executing the experiment automatically by using the multi-armed bandit method to minimize costs; d. allowing the user(s) to view results as they are received and updating the probability of heuristic success or failure.
15. The system of Claim 14, wherein the method may be implemented in any suitable hardware, software, firmware, or combination thereof.
16. A system adapted for forecasting real world sales of products, comprising; a. a computing system for forecasting retail sales of one or more products, wherein the computing system further includes a hardware processor and a memory coupled to the hardware processor; b. plurality of subsystems consisting of;c. a data receiver subsystem configured to receive one or more inputs associated with uniqueness of the one or more products; d. an availability determination subsystem communicatively coupled to the data receiver subsystem to determine availability of the one or more products in market based on the received one or more inputs; e. a data obtaining subsystem coupled to the availability determination subsystem to obtain one or more products state based on the determined availability of the one or more products; wherein one or more equations are generated to determine the current state of the one or more products; f. a data generation subsystem communicatively coupled to the availability determination subsystem and the data obtaining subsystem, wherein the subsystem generates one or more time series components based on the determined availability of the one or more products; g. a data forecasting subsystem coupled to data obtaining subsystem and the data generation subsystem to forecast retail sales of the one or more products based on the generated one or more time series components using data forecasting methods based on Artificial Intelligence models.
17. The system of Claim 16, wherein the machine learning modules include Croston and EWMA are applied to the obtained one or more products state.
18. The system of Claim 16, wherein the system further includes a product determination subsystem to determine the one or more products required to be placed on each of the one or more stores based on the forecast retail sales of the one or more products and one or more substitution models.
19. The system of Claim 16, wherein the computing system incorporates one or more attributes to enrich generated data by time series components and the plurality of one or more products state.
20. The system of Claim 19, wherein one or more attributes comprises product attributes generated by Natural Language Processing (NLP) tokenization, store attributes generated by geographical vectors, competition vectors, weather-based attributes and customer personas.
21. A method for forecasting retail sales of products using system as claimed in Claim 16, comprising the steps of; a. receiving data of the one or more products and determining its uniqueness by the data receiving subsystem; b. determining the current availability of the one or more products at each of the one or more stores by the availability determination subsystem; c. generating equations to determine the current state of the one or more products currently available by the data obtaining subsystem; d. determining whether the one or more stores are producing intermittent sales or regular sales of the one or more products; e. applying Croston forecasting model and EWMA forecasting model for forecasting the retail sales of the one or more products by the data forecasting subsystem; f. generating time series components by the data generation subsystem;g. enhancing the output in Step f by enriching it with other attributes comprising product attributes, customer personas by the data generation subsystem ; h. forecasting the retail sales of the one or more products at each of the one or more stores by the data forecasting subsystem upon receiving the output from the Steps e and f; and i. determining the adjustments required to availability of the one or more products at each of the one or more stores in accordance with the forecasted retail sales of the one or more products.
Citation Information
Patent Citations
Method and apparatus for optimizing a multi-dimensional space
US20150199627A1
Commerce System and Method of Providing Shopping Agents and Sales Agents for Managing Purchasing Decisions
US20150324882A1
Product correlation analysis using deep learning
US20190347611A1
System and method for cognitive adjacency planning and cognitive planogram design
US20200034781A1
Distributed Ledger for Additive Manufacturing in Value Chain Networks
US20220197247A1
