Dear all

I have a database in which for each day I have difference variables.I need to discover patterns and rules.

Example:

I have for each day a basket of products (bread, ham, milk,honey,pasta).

Different phenomena can take place both in the same day (bread & ham are bought together or when I take pasta, I don't take bread) and this could be found with correlation analysis.

It is also possible that an event in the past can affect a decision in the present. For example...If yesterday I took pasta, today I will not buy pasta again...A value of the past is correlated with a value in the present (the two values can also belong to different variables: bread and pasta)...Let's call them dynamic patters (if you have a more suitable definition, please tell me)

The question is: do you have a methodology to discover dynamic patterns in a timeseries database?

Thank you very much

Dario

PS:In attachment an example how database looks like.

Similar questions and discussions