New Product Demand Prediction Using Similar Product Sales Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing demand prediction techniques struggle to accurately forecast long-term sales of new products after their promotion period, lacking sufficient data for reliable predictions.
Innovation Solution
A new product demand prediction apparatus and method that utilizes information from similar products based on promotion scale, characteristics, and sales methods to predict sales in a predetermined period post-promotion, incorporating first and second similar product information for accurate long-term forecasting.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If demand prediction is performed for new products after promotion period using existing techniques, then prediction can be made, but prediction accuracy deteriorates due to lack of sufficient learning data
Solution Approach 1:
The patent creates virtual learning data by copying and adapting sales data from similar products (both within and outside the category) to serve as training data for the new product's demand prediction model, resolving the data scarcity issue while maintaining prediction accuracy
Solution Approach 2:
The patent introduces similar products as intermediary data sources that bridge the gap between available historical data and the needed prediction data for new products, enabling accurate long-term forecasts without direct learning data from the target product
2Measurement precision
If demand prediction focuses on short-term sales during promotion period, then prediction accuracy is improved, but long-term demand prediction capability deteriorates
Solution Approach 1:
The patent extends prediction from the short-term promotion period dimension to the long-term post-promotion period dimension by incorporating similar product data that spans multiple time periods, enabling accurate long-term forecasting without sacrificing accuracy
Data Source
AI summary
A new product demand prediction apparatus according to the present disclosure includes: a memory configured to store instructions; and one or more processors configured to execute the instructions to: acquire new product information including at least information regarding promotion of a new product that is a target of demand prediction; acquire first similar product information indicating a sales performance of a first similar product selected based on at least one element of a scale of the promotion, a characteristic of a product, and a sales method; acquire second similar product information indicating a sales performance of a second similar product selected based on the category of the product; and predict sales in a predetermined period including a period after a promotion period of the new product based on the new product information, the first similar product information, and the second similar product information.


