User Purchase Limit Prediction with Multi-Branch Regression

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing neural network models for purchase prediction on users consider only a single factor, leading to inaccurate final purchase limits due to insufficient comprehensive consideration of factors.

Innovation Solution

A method involving a target model that performs multiple types of regression processing, including feature extraction and multiple regression branches, to obtain comprehensive purchase probabilities and limits, enhancing accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a neural network model uses only a single regression processing to determine the final purchase limit, then the model structure is simple, but the prediction accuracy is insufficient

Engineering Contradiction:
Improveprediction accuracyVSAvoidmodel structure complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent divides the prediction task into multiple independent regression branches, each handling different aspects of purchase limit prediction. This segmentation allows each branch to specialize in specific factors (e.g., user behavior patterns, product characteristics, contextual information), improving overall prediction accuracy while maintaining manageable model complexity through modular architecture

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent implements a multi-functional regression model that simultaneously performs multiple types of regression processing (e.g., linear regression, logistic regression, decision tree regression) within a single unified model structure. This multi-functionality enables the model to consider diverse factors and prediction approaches concurrently, enhancing prediction accuracy without requiring separate models for each factor

Inventive Principle:
Principle #6Universality (Multi-functionality)

2Measurement precision

If the model considers only one factor for purchase limit determination, then the model is simple and easy to implement, but the prediction accuracy is not high enough

Engineering Contradiction:
Improveprediction accuracyVSAvoidfactor consideration comprehensiveness
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent transitions from single-factor prediction to multi-dimensional prediction by introducing multiple regression branches that each process different dimensions of data (e.g., temporal patterns, spatial relationships, categorical attributes, continuous variables). This dimensional expansion allows the model to capture complex interactions between factors, significantly improving prediction accuracy while maintaining adaptability through a systematic multi-dimensional framework

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

Data Source

PatentEP4685723A1Consumption prediction method and related device
Publication Date: 2026.01.28 HUAWEI TECH CO LTD
  • EP4685723A1 patent drawingFigure 1
  • EP4685723A1 patent drawingFigure 2a
  • EP4685723A1 patent drawingFigure 2b

AI summary

A purchase prediction method and a related device thereof are provided. The method includes: obtaining information associated with a user (501); performing feature extraction processing on the information, to obtain a purchase probability of the user (502); performing first regression processing on the information, to obtain a first purchase limit of the user (503), and performing second regression processing on the information, to obtain a second purchase limit of the user (504), where the first regression processing and the second regression processing are different regression processing; and finally obtaining a predicted purchase limit of the user based on the purchase probability, the first purchase limit, and the second purchase limit (505).