Segmentation System Using Supervised Learning for Conversion Potential Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional segmentation systems fail to create clusters with varying conversion potentials, often resulting in groups with similar conversion potentials and limited size due to the indirect relationship between group size and conversion potential.
Innovation Solution
A supervised learning method combined with clustering analysis is used to identify superior clusters by training a model with user features and historical conversion data, selecting a subset of high-ranked user features, and performing hierarchical clustering analysis to segment populations into clusters with varying conversion potentials.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional unsupervised grouping methods are used to segment populations based on common features, then groups are formed with similar conversion potentials, but the conversion potential cannot be differentiated or improved across groups
Solution Approach 1:
The patent changes the parameter used for grouping from common features to conversion potential metrics. By using supervised learning to identify features correlated with conversions and ranking them by importance, the system transforms the grouping criterion to directly target conversion potential differentiation, resolving the contradiction between measurement precision and operational simplicity
Solution Approach 2:
The patent replaces the mechanical/conventional unsupervised grouping method with a data-driven supervised learning approach. Instead of manually defining groups based on common features, the system uses machine learning models to automatically identify and rank features by their correlation with conversions, substituting traditional segmentation mechanics with intelligent algorithms
2Measurement precision
If population is segmented into smaller groups based on individual conversion history, then higher individual conversion potential is achieved, but the overall group conversion potential remains limited due to randomization of features
Solution Approach 1:
The patent changes the grouping parameter from individual conversion history to ranked feature subsets. By selecting top-ranked features that are strongly correlated with conversions and using these as the basis for clustering, the system maintains high conversion potential while enabling larger group sizes, as members share common high-value features rather than requiring identical conversion histories
Solution Approach 2:
The patent applies segmentation by dividing the population into clusters based on shared high-ranked features. This creates intermediate groups that are larger than individual-based segments but more homogeneous in conversion potential than conventional unsupervised groups, resolving the contradiction between group size and conversion potential
3Adaptability or versatility
If all user features are used in clustering analysis, then comprehensive grouping is achieved, but conversion potential varies less across clusters due to inclusion of low-correlation features
Solution Approach 1:
The patent extracts only the most important features by ranking all user features based on their correlation with conversions and selecting the top-ranked subset. This extraction process removes low-correlation features that would dilute conversion potential variation, while retaining comprehensive coverage of high-value features that drive conversion differences across clusters
Solution Approach 2:
The patent applies local quality by assigning different weights to different features based on their importance. Instead of treating all features equally, the system identifies and emphasizes locally important features (those with high conversion correlation) while downplaying or excluding less important features, creating clusters with more meaningful conversion potential differentiation
Data Source
AI summary
A segmentation system utilizes a supervised learning method and a clustering analysis to identify clusters, thereby segmenting a population into groups, where the clusters are associated with various conversion potentials that indicate the probability of an event. The segmentation system employs the supervised learning method to train a model on training data comprising historical conversion data and features associated with members of the group. A subset of features is selected from a ranked order that is determined using weights generated by the supervised learning. A clustering analysis is performed for a population with respect to the subset to generate clusters. A superior cluster is identified based on it having a conversion potential greater than a conversion potential of another cluster. In a marketing context, the system can be employed to identify a superior cluster of users that have a higher conversion potential in response to an advertisement campaign.