2

I am building a model of customer spending, and need some help to identify the best way of doing this.

(1) I have an exponential distribution for each customer's spending (2) Bernoulli distribution for whether or not that customer comes at all

For each marketing campaign (a couple of thousand customers) we combine all the cases of [(1)and(2)]. Then we combine all the marketing campaigns for one year. I'm sure this is a common scenario, but I cannot find a definitive answer for what distributions result from the combinations: [Exp+Bernoulli] -> add up for one campaign -> [distribution?] -> add up all for the year -> [distribution?]

Would appreciate it if someone can confirm for me the theory on this, and also any references to how this is implemented in code!

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
Snow64
  • 29
  • 1
  • 1
    It isn't clear that you really want a (single) distribution for the combination. You may want to explore a *hurdle* model (df. [What is the difference between zero-inflated and hurdle distributions (models)?](http://stats.stackexchange.com/q/81457/7290)) – gung - Reinstate Monica Feb 24 '16 at 23:13
  • Why do you want to *add* them? It would seem more natural to multiply. Can you explain? – kjetil b halvorsen May 05 '19 at 09:34
  • 2
    I'm voting to close this question as off-topic because cannot be answered without clarifications, and the OP is gone. – kjetil b halvorsen May 05 '19 at 09:37

0 Answers0