Predicting likelihood of attending next event I work in a sports stadium and I would like to run a predictive model on all people who have previously bought tickets to events, to predict the likelihood of them returning to the next event, or indeed any event in the future. 
We have hundreds and thousands of customers, and approximately 40 historical events in which we can use as binary data on whether or not they attended each event.
Does anyone know of a method I can use to run such  a predictive model? I am not really sure where to start, although I believe this can be done.
Thanks very much in advance.
 A: To elaborate on @Tony-Frame 's answer, you have to decide on what to put on the "right-hand side of the equation". Do you have any individual explanatory variables, like socio-economic data on the customers? In case you don't, can you characterize and categorize these 40 events somehow (in terms of "general popularity", intensity of marketing, things like that? These could be also included in the specification in the right-hand side, as common regressors to all equations. 
If no potential regressors are available, then you can create an autoregressive model as presented in my answer to this post : logistic regression model for process with memory 
This would mean, predicting each customer "separately" without exploiting any possible interdependence on how customers decide... ("he goes -so I'll go"). Trying to taking into account this interdependence, would lead you into a system of equations and panel-data logistic regression.
A: Well, without knowing the data you have, the first suggestion I can make is that you will want to build a logistic regression. You can read more about the definition here, but since you are looking to compute likelihoods/probabilities, logistic regression is very well suited for that purpose. And you're on the right track thinking you can use historical attendance as the binary response variable for building the model. What you will want to do is look at the attributes you have to see which will be most indicative of attendance or repeat attendance.
Hope this helps.
