I do not have an actual link but I can point out a book that covers the basics of AI at a level that is, in my opinion, easier to digest compared with other fundamental AI books. It is Patrick Winston (from MIT) and the book is called Artificial Intelligence. It goes over all different AI techniques and what I was referring to would be i…
I do not have an actual link but I can point out a book that covers the basics of AI at a level that is, in my opinion, easier to digest compared with other fundamental AI books. It is Patrick Winston (from MIT) and the book is called Artificial Intelligence. It goes over all different AI techniques and what I was referring to would be in the inductive learning category. In short, what they probably have is some basic set of "instances" that rather than being geographically focused, they are more likely to be feature of interest based - as in regularly reads or watches sources X, Y and Z, buys consistently A, B and C and so on. Then they generalize over the instances which would have substantial variations and get a "class" that can be used to predict the main "properties" that define the person - for example, all "troublemakers" will read source X and W while buying items B and F. This can be extended and made more detailed if needed - if there are any geographic based patterns that can be derived using the class definition. I am not sure if that makes any sense but that I the quickest way I can summarize one basic way to do the analytics.
I do not have an actual link but I can point out a book that covers the basics of AI at a level that is, in my opinion, easier to digest compared with other fundamental AI books. It is Patrick Winston (from MIT) and the book is called Artificial Intelligence. It goes over all different AI techniques and what I was referring to would be in the inductive learning category. In short, what they probably have is some basic set of "instances" that rather than being geographically focused, they are more likely to be feature of interest based - as in regularly reads or watches sources X, Y and Z, buys consistently A, B and C and so on. Then they generalize over the instances which would have substantial variations and get a "class" that can be used to predict the main "properties" that define the person - for example, all "troublemakers" will read source X and W while buying items B and F. This can be extended and made more detailed if needed - if there are any geographic based patterns that can be derived using the class definition. I am not sure if that makes any sense but that I the quickest way I can summarize one basic way to do the analytics.