I am trying to calculate a metric "frequency of purchase of a customer" to determine when I would expect a customer to then purchase an item next.
Say, the distribution of my customer base looks like that in the images below (it is truncated at about 70 days but you get the idea) additionally, '-1' refers to a transaction not being followed by another. I then want to determine for a group of customers based on how many days ago they last purchased, how many I expect to purchase within the next n days. How would I go about doing this?
To make this question less open-ended, how would I calculate the likelihood that a customer who last spent 5 days ago will spend within the next 10 days?