Balance Prediction product is available as a Private Beta version.
Yapily’s Balance Prediction assesses previous transaction data to forecast the future balance of a business or individual user, to provide confidence in their available funds.
Balance Prediction analyses historical transaction data for the past 60 days, and uses this to model future account balance for the next 30 days, using data from a user's accounts, across multiple banks. Using models based upon spending patterns over several months, the median predicted balance is returned alongside the 10th and 90th percentile to give additional confidence that the value will be within this range.
This information can help you understand the future pattern of your customer’s financial behaviour and trending data over time.
Balance Prediction is calculated using a time-series implementation model that uses root mean square error as its loss function.
It is expected that Balance Prediction is run on a primary spending account. As a result, the training data only contains accounts with a transaction history of at least 3 months and 200 total transactions.