Merchant Classification via Dynamic Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional merchant classification methods are slow, complex, and fail to consider financial and operational attributes of each merchant, often leading to incorrect merchant hierarchies and classifications due to reliance on static rules and incorrect tagging by acquiring banks.
Innovation Solution
A computer-implemented method using a server system that accesses a historical transaction dataset to generate key performance features for each merchant, determines hyper-parameters for a clustering machine learning model, and labels merchant clusters into different classes based on a classification threshold.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional merchant classification methods are used, then the process is simple to implement, but the classification accuracy is low and the process is slow
Solution Approach 1:
The patent replaces the conventional mechanical/rules-based classification system with a machine learning-based system. Specifically, it uses a clustering algorithm (DBSCAN) that automatically learns merchant categories from transaction data without relying on pre-defined rules or hierarchies. This substitution enables the system to achieve high classification accuracy by detecting inherent patterns in the data rather than relying on manually constructed classification rules.
Solution Approach 2:
The system performs self-service by automatically determining merchant categories through unsupervised clustering. The algorithm autonomously identifies merchant groups based on transaction characteristics without requiring manual intervention or pre-labeled data. The clustering process self-adjusts to the data distribution, automatically discovering optimal category boundaries and merchant groupings based on the inherent structure of the transaction data.
2Adaptability or versatility
If static predefined rules are used for classification, then the system is easy to operate, but the results are static and fail to provide actionable insights
Solution Approach 1:
The patent implements dynamics by replacing static classification rules with a dynamic machine learning model that adapts to changing financial patterns. The clustering algorithm continuously processes new transaction data, allowing merchant category definitions to evolve automatically in response to changing economic conditions, consumer behavior, and market trends. This enables the system to maintain relevance and provide actionable insights in a dynamic financial environment.
Solution Approach 2:
The system incorporates feedback mechanisms where the clustering results can be validated and refined through continuous monitoring of transaction patterns. The model learns from actual merchant behavior and transaction data, adjusting category assignments over time. This feedback loop enables the system to correct misclassifications and adapt to emerging merchant types or category boundaries, providing increasingly accurate and actionable classification results.
3Reliability
If reliance on acquiring bank data is used, then the data source is readily available, but incorrect tagging leads to formation of incorrect merchant hierarchy
Solution Approach 1:
The patent introduces an intermediary processing layer between the raw acquiring bank data and the final merchant classification. The clustering algorithm acts as this intermediary, cleaning, validating, and reorganizing the data to eliminate incorrect tagging and hierarchical structures. This intermediary step transforms potentially inaccurate bank-provided classifications into more reliable merchant groupings by independently analyzing transaction patterns and identifying true merchant relationships.
Solution Approach 2:
The system substitutes the mechanical data collection approach (relying on bank-provided merchant codes) with an automated machine learning process that independently verifies and corrects merchant classifications. The clustering algorithm analyzes multiple transaction attributes simultaneously, enabling it to detect and correct incorrect tagging by acquiring banks, thereby improving the reliability of merchant hierarchy formation.
Data Source
AI summary
Methods and server systems for performing merchant classification are described herein. The method performed by a server system includes accessing a historical transaction dataset including transaction attributes from a database. Method includes generating a set of key performance features for each merchant of the plurality of merchants based on the historical transaction dataset. Method includes determining a set of hyper-parameters including an epsilon value and a Minimum points (MinPts) value for a clustering machine learning (ML) model based on a K-nearest neighbor (KNN) plot generated based on the set of key performance features for each merchant. Method includes generating via the clustering ML model, a set of merchant clusters based on the set of key performance features for each merchant and the set of hyper-parameters. Method includes labeling each merchant cluster as one of a first merchant class and a second merchant class based on a classification threshold.