Categorization and Clustering PRO
Credit Insights leverages Open Banking data to give a detailed budget analysis of your users. It is composed of a categorization and clustering engine that label transactions and group them together.
Categorization
Our categorization engine takes transaction descriptions and applies a category field and a type field to each one.
The category describes the purpose of one transaction and attempts to ascertain the reason why the user made it. Our Category taxonomy is specialized in risk assessment and credit worthiness analysis. It can be split into 2 groups :
- Entries (revenue, new financing, shareholder incoming, ...)
- Expenses (loan repayments, overdraft fees, contribution paid to URSSAF, ...)
The type on the other hand describes the means of how the transaction was executed. Algoan taxonomy contains 13 types such as payment by card, cash deposit, direct debit, and check. See Taxonomy page to grasp a full view.
When a category or a type cannot be determined, Credit Insights sets the category field to either OTHER or, OPEX_OTHER depending on the sign of amount.
Clustering
The clustering algorithms aims at automatically grouping transactions in order to reproduce a human analysis of a bank statement.
Once transactions are categorized, the clustering process enables Credit Insights to identify specific budget lines. Transactions that share similarities are packed together into a cashflow. A cashflow is the base block of most of the Credit Insights indicators.
Rejected Transactions
Rejected transactions appear twice in one bank statement. The first one is related to the debit operation and can have any category and the credit operation can have either RED_FLAG_REJECTION_ON_PAYMENT or RED_FLAG_REJECTION_ON_LOAN_REPAYMENT. The Algoan clustering engine is able to link the credit transaction with the debit transaction so that your service can know exactly which transaction was rejected.
This information is detailed in enrichments.relatedTransactions in the Transaction model.