This is a simple Kotlin application that leverages Naive Bayes or logistic regression to categorize bank transactions.
NOTE: Kotlin-Statistics now has this feature in an easy-to-use API.
Input and categorize a few transactions, and the simple AI will start to predict the categories once it picks up probabilities for given keywords.
Chapter 13 of O'Reilly Data Science From Scratch by Joel Grus was a useful reference.
Here is a recorded demo. The first three transactions did not have enough data to predict. But after the fourth transaction, the algorithm starts to make accurate predictions.