Demand Forecasting Model Using Intent-Demand Lag Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional supply chain management and marketing dashboards struggle to accurately forecast customer demand due to the lag between customer intent and demand, leading to inefficiencies and inaccurate financial planning, as existing ML algorithms often overfit or underfit training data, failing to capture complex demand patterns and anomalies.
Innovation Solution
A system that utilizes both customer digital intent data and demand data to train and test prediction models, employing a combination of ML algorithms to determine the relationship between intent and demand, including feature engineering, ensemble learning, and recursive learning to generate an optimal prediction model, which is then used to create a forecast index for dynamic supply chain adjustments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional ML algorithms are used to forecast customer demand, then prediction models can be generated, but the models tend to overfit or underfit training data resulting in inaccurate predictions
Solution Approach 1:
The patent divides the prediction model into multiple specialized components: a customer intent prediction model that processes digital intent data, a demand prediction model that processes historical demand data, and a lag determination model that identifies time lags between intent and demand. This segmentation allows each model to specialize in specific patterns, reducing overfitting while maintaining accuracy.
Solution Approach 2:
The patent merges the outputs of multiple independent models (intent prediction model, demand prediction model, lag determination model) into a unified forecast through an ensemble approach. This combination leverages the strengths of each specialized model while canceling out individual weaknesses, improving overall prediction reliability.
2Extent of automation
If ML models are trained with sampled data to make predictions, then the models can forecast without explicit programming, but generating accurate prediction models remains difficult
Solution Approach 1:
The patent implements feedback mechanisms where the lag determination model continuously refines its estimates based on the relationship between intent data and actual demand outcomes. This feedback loop allows the automated model to learn from prediction errors and improve accuracy over time without requiring reprogramming.
Solution Approach 2:
The patent dynamically adjusts key parameters including time lags, weighting factors for different data sources, and model complexity based on the specific characteristics of the training data. This adaptive parameter adjustment enables the automated model to optimize accuracy for different business contexts without manual intervention.
3Productivity
If supply chain management uses traditional forecasting methods, then processes can be simplified, but customer demand cannot be accurately forecasted leading to inefficiencies and incorrect financial plans
Solution Approach 1:
The patent introduces temporal dimensionality by explicitly modeling time lags between customer intent and demand realization. It also adds a multi-data-source dimension by integrating both digital intent data and historical demand data. This dimensional expansion enables accurate forecasting while maintaining computational efficiency through structured data processing.
Data Source
AI summary
The disclosure includes a system and methods of generating a forecast index based on customer digital intent data. The methods include generating a plurality of prediction models using a unique machine learning algorithm for each prediction model, determining an optimal prediction model based on an overall model score assigned to each model of the plurality of prediction models, and generating a forecast index based on the optimal prediction model. The forecast index can be used to adjusts the supply of one or more products based on predicted customer demand.