Identify income for customers with regular or irregular income sources, including part-time and freelance jobs.
Identify all recently received loans and loan repayments. Ideal in cases where full credit history is not available.
Optimised for Risk.
Our categorisation engine is built to recognise the purpose of a transaction based on its description. The engine was built to replace manual bank statement analysis by credit specialists as well as improve risk-critical areas, where traditional Personal Finance Management (PFM) categorisation engines underperform—it was built with clear focus on risk and financial behaviour.
How categorisation works?
Nordigen engine uses the transaction's details field to match keywords, phrases and patterns and identify category of the payment or transfer.
Extracting categories from raw transaction data can be valuable across departments. This means being able to get "missing" data for real time scoring, pre-scoring as well as for customer segmentation or analytics.
Simple API, supports multiple file formats
Our RESTful API supports a wide array of file formats, including our own JSON format, as well as transaction files (JSON, XML) directly from account aggregators such as Yodlee, Instantor, Kontomatik, Pich Technologies, Saltedge as well as bank statements in PDF formats in selected countries (e.g. Latvia, Lithuania, Estonia, Finland).
Hosted on the secure Amazon Web Services cloud
Security and scalability are our top priorities. For this reason we've deployed our solution on AWS (servers located in Dublin, Ireland) to ensure that your data processed securely and the service is always running.
Multiple geographies and languages supported
Our categorisation engine currently supports transactions from Spain, Germany, Poland, Czech Republic, Sweden, Denmark, Finland, Latvia, Lithuania, Estonia, Australia and New Zealand.
Next (in 2018): UK, Mexico, Philippines, Brazil, South Africa
Identify 150+ categories in every statement
We have created an extensive category tree of income and expense categories. This gives a lot of freedom to data science and credit modelling teams to build inventive decision rules an tod identify financial behaviours previously hidden within the data.
High categorisation rate, continuously learning
To deliver the best results, we employ multiple keyword matching techniques alongside prediction algorithms and continuously teach the engine. We employ a number of internal data operators to label transactions, test the categorised data as well as spot and fix false-positives. The best part, however, is the fact that we work with multiple lenders in any country where we operate, which allows us to learn from multiple parties at the same time and make improvements to all our customers.
We measure our success in categorisation rate (% of transactions that are identified and not fall into category "Other") and error rate (% of transactions that were categorised incorrectly).
- average categorisation rate across our 7 core markets: 80%
- highest categorisation rate: 92% (Latvia, Nov 2017)
- estimated average error rate across our 7 core markets: below 0.5%
Nordigen categorisation API supports different types of input files and formats, including JSON and XML files from the most popular account aggregators and banking APIs. The example below reveals how the categorisation works in real life and what is the expected output.
Input file (JSON)
Output file (JSON) with categories
Interested to see a live demo?
We already receive categories from our account aggregator. Why choose Nordigen on top?
Higher overall accuracy. Both solutions are complimentary and higher quality data means better decisions.
Is there a way to use the categorisation engine without integrating to the API?
Yes, we have a browser-based version of the engine that works exactly like the API, but shows the categorisation results visually. In fact, this video shows how the browser-based solution works. Just contact our sales and let us know about your use-case.