Hi Monzo team and community.
Came across an interesting one today which I assume was down to the Monzo machine learning algo not being quite clever enough (or not having enough data).
We’ve had our joint account with Monzo now for 2.5 months and our mortgage is on DD to come out on 1st of each month.
So far so good and works as intended.
In June this year, the 1st was a Saturday, so it was going to take it on the 3rd which was the Monday. Monzo app told me this with no issues in my forthcoming spending summary.
Today our mortgage came out on the 1st July as expected but in our committed summary spending it is showing that it has gone today and also due to leave on the 3rd July.
From what I can ascertain, the predictive model that picks up past usage for committed spending has seen that we paid this amount to this organisation last month on the 3rd so it should be coming again on the 3rd (it won’t, I checked).
Does this model train with experience i.e. if I have paid it on the 1st for 9 months out of 12 and the outliers being the 2nd, 3rd, 4th or 5th (the 1st having fallen on a weekend or a bank holiday), will it eventually learn that this payment will come out on the 1st even if the month before it was paid slightly later as per this scenario?
Only way I can see to test without bankrupting myself is to wait for the next time this would occur should my analysis be correct. This is in October where in September this payment will be on the 2nd due to a weekend and the October 1st is on a Tuesday. If my analysis is correct, in my spending summary on Oct 1st I would expect to have seen my mortgage payment left and also one in the committed spending summary for the 2nd October.
Any insight would be great. If there is an opportunity here to update the predictive models in the app then even better!
Thanks
Andy