This data is from a public data source but is very similar to the data we usually see. It is 80 000 transactions over a 7-year period.

The ID column on the sample submission is created from DATE and BALANCE AMT columns in the test; to make it easier for the hack we have also provided the ID column in the test set, you will notice that this is not provided in the train set; since this is not a feature to use in your modelling.