Inventory Anomaly Prediction Model for Auditing Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing inventory management systems face challenges in detecting and correcting inventory anomalies, such as 'pencil whipping' or 'gaming', which lead to inaccurate data that affect supplier forecasting and other business decisions, with current methods being costly and time-consuming.
Innovation Solution
A computerized method using a two-stage machine learning process involving unsupervised learning to generate labeled training data and a stacked ensemble of logistic regression, random forest, and decision tree processes to train an anomaly prediction model, which identifies and corrects inventory anomalies by scheduling audits or notifications.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional inventory auditing methods are used to detect anomalies, then detection reliability is improved, but productivity deteriorates due to excessive workload and time consumption
Solution Approach 1:
The system performs preliminary anomaly detection using machine learning models before full auditing. The anomaly prediction model identifies high-risk inventory items in advance, allowing auditors to focus only on predicted anomalies rather than conducting comprehensive audits of all items, thus maintaining detection reliability while improving productivity
Solution Approach 2:
The patent replaces manual mechanical auditing with an automated machine learning-based anomaly detection system. The model automatically analyzes inventory data patterns, associate behaviors, and historical records to identify anomalies, substituting human labor with computational analysis to improve both efficiency and consistency
2Measurement precision
If comprehensive inventory auditing is performed to ensure data accuracy, then measurement precision is improved, but loss of time increases
Solution Approach 1:
The system applies different auditing intensities to different inventory items based on their anomaly risk scores. High-risk items identified by the model receive thorough auditing attention, while low-risk items undergo minimal or no auditing. This localized quality approach ensures measurement precision for critical items while reducing overall time loss
3Reliability
If frequent inventory audits are conducted to maintain data accuracy, then reliability is improved, but productivity deteriorates due to increased workload
Solution Approach 1:
The system implements periodic anomaly detection using the trained model at optimized intervals rather than continuous auditing. The model predicts anomalies based on accumulated data patterns, allowing the system to audit only when predicted anomalies are likely, maintaining reliability while reducing the frequency of disruptive audits and preserving operational productivity
Data Source
AI summary
A computerized method trains an inventory anomaly prediction model and uses the model to predict and address inventory anomalies. Daily inventory data associated with an inventory storage location is obtained from an inventory data store and provided to a pretrained anomaly prediction model as input. The anomaly prediction model generates an anomaly indicator associated with an item type in the obtained inventory data. Corrective action is then caused to be performed in association with the item type and based on the anomaly indicator, whereby inaccuracy of an inventory count of the item type in the inventory data store is corrected. Further, in some examples, the anomaly prediction model is trained using machine learning techniques and training data that is generated and labeled using a tuned outlier detection model. Thus, the method enables the anomaly prediction model to be efficiently trained for accuracy using the described two-stage training process.