Historical & Live Data
Historical and live Data​
Historical data​
To improve our fraud detection models and accurately identify fraud cases, we require at least 24 months of historical data consisting of both payment transactions and chargebacks. We use this data to train our models and improve performance, enabling us to develop highly customized fraud detection models that are specifically tailored to your business.
By training our models on your specific data, rather than generic data, we can optimise our fraud detection and prevention measures to provide more accurate and effective results. This helps to reduce the risk of fraudulent transactions and minimize potential losses for your business.
To ensure the highest level of accuracy, we recommend minimizing the time gap between sending your historical data and starting to send live transactions to the fraud score endpoint. This helps to avoid gaps in the timeline that could impact the accuracy of our fraud scoring models, ensuring that you receive the most effective protection against fraud.
Live data​
To score your transactions and effectively use our fraud detection services, you'll need to send your live transaction data to the fraud score endpoint using the API. These requests should occur pre-authorization and require accurate mapping of your data fields and values to the equivalent Fraudio data fields and values. Visit our FAQs for more information on pre- and post-authorization data.
To optimise our fraud detection and prevention measures, we require both transaction data and batch data. When going live, your transaction data will need to be processed in real-time or sent in batch on a frequent basis (we recommend every 12 to 24 hours) to ensure accuracy and timeliness. The batch data can be updated less frequently - for example, on a monthly basis.
How to transfer historical data to Fraudio?​
To transfer historical data to Fraudio, simply upload CSV files to our dedicated storage on Amazon S3. We've chosen this storage solution for its wide usage, easy integrations, and compliance with security and privacy regulations.
To upload files, we'll securely share a key ID and secret with you upon request. Many S3 clients are available for uploading, including the command-line interface, various GUIs, and client-side libraries for most programming languages. For more detailed information, please refer to our batch transfer integration manual.
If you prefer alternative methods for sending your data, please contact us to discuss. And remember, the more data we receive, the better your fraud scores will be!