Dynamic Minimum Presentation Engine for Fresh Item Production Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems for predicting fresh item production face challenges in accurately determining the optimal quantity to produce daily, leading to issues of overproduction and underproduction, which result in financial losses and customer dissatisfaction.
Innovation Solution
A system that uses dynamic data and machine learning to calculate a dynamic minimum presentation (MP) value, taking into account item attribute data, such as shelf life and production cost, to optimize safety stock and minimize monetary losses associated with overproduction and underproduction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If rules-based systems are used to predict fresh item production quantities, then the system is simple to implement, but prediction accuracy is poor leading to lost revenue and increased waste
Solution Approach 1:
The patent replaces traditional rules-based mechanical prediction systems with a machine learning model that uses historical data and dynamic factors to generate demand probability distributions. This substitution enables significantly improved prediction accuracy by capturing complex patterns in demand that rules-based systems cannot detect, while the model can be deployed as a software service that integrates seamlessly into existing production planning workflows.
Solution Approach 2:
The patent transforms the prediction approach by changing from deterministic rules to probabilistic predictions. The machine learning model outputs a demand probability distribution rather than a single fixed value, allowing the system to account for uncertainty and variability in demand. This parameter change enables more flexible production decisions that can adapt to different risk tolerances and business objectives.
2Reliability
If more fresh items are produced to prevent stockouts, then customer satisfaction improves, but monetary loss from unsold items increases
Solution Approach 1:
The patent changes the production decision parameter from a fixed safety stock level to a dynamic quantity derived from the demand probability distribution. By calculating the optimal production quantity as the sum of predicted demand and a data-driven safety stock component, the system achieves the right balance between preventing stockouts and minimizing overproduction waste, adapting to current demand conditions rather than using static thresholds.
Solution Approach 2:
The system incorporates feedback loops where actual sales data and demand patterns are continuously fed back into the machine learning model to refine future predictions. This feedback mechanism allows the system to learn from past performance and progressively improve its ability to balance availability and waste reduction, adjusting production recommendations based on real-world outcomes.
3Loss of substance
If fewer fresh items are produced to reduce waste, then monetary loss from unsold items decreases, but the risk of stockouts and lost sales increases
Solution Approach 1:
The patent transforms the safety stock determination from a conservative fixed value to a dynamic calculation based on demand uncertainty. The optimal production quantity formula explicitly incorporates the safety stock component derived from the demand probability distribution, allowing the system to maintain adequate availability buffers only when the probability distribution indicates higher uncertainty or demand variability, rather than applying uniform conservative buffers to all situations.
4Adaptability or versatility
If static safety stock levels are used to prevent stockouts, then the system is easy to manage, but it cannot adapt to varying demand patterns leading to inefficiency
Solution Approach 1:
The patent implements dynamics by replacing static safety stock levels with a dynamic calculation that adapts to varying demand patterns. The machine learning model generates demand probability distributions that automatically adjust to changing demand conditions, and the optimal production quantity is recalculated based on current demand characteristics rather than fixed historical averages. This dynamic approach allows the system to automatically adapt to seasonal variations, promotional effects, and other demand drivers without manual intervention.
Solution Approach 2:
The system performs self-service by automatically generating production recommendations without requiring manual safety stock setting or complex planning procedures. The machine learning model and optimization algorithm work autonomously to determine optimal production quantities based on historical data and current demand patterns, reducing the complexity of production planning while improving adaptability to changing conditions.
Data Source
AI summary
Examples provide for generating dynamic minimum presentation (MP) values for optimizing fresh item production while reducing system resource usage. An MP engine calculates a monetary level impact of overproduction and underproduction of a fresh item using production-related data, including a shelf life of the fresh item and cost of production. The calculated monetary level impact includes predicted monetary loss from lost sales due to an item going out-of-stock and/or monetary loss from unsold instances of the fresh item where too many instances of the fresh item are prepared. The dynamic MP value is generated based on dynamic data, the item attribute data, and the predicted monetary loss. The dynamic MP value is used to create a production plan recommendation customized at a store-item level for a selected date, thereby enabling optimization of safety stock while minimizing system resource usage consumed in generating MP predictions for fresh items.