Please advise on methods and models for prediction of future events sequenced in time based on some previous event history.

I need to solve the following problem:

Input data: a huge (750 GB) set of records with user reading history, where record has the following structure:

userId, title, author, genre, date_read

All titles depict a wide sellection of fiction including modern, old, popular and classics.

Task: Given partial reading history of a user predict what next titles and in what oreder she will most probably read? To be more precise: system must predict an ordered sequence of next titles for a user who has read just a few ones from all available titles in the data set. In other words predict ordered sequence of titles for a user who just started reading. It is importatnt to predict not just what books user will read, but also the in what order user will read these titles. So in fact the task is to determin how reader tastes evolve over time.

It looks like time series problem that would require Markov Chain model. On the other hand data of similar-minded readers may be used in prediction, so collaborative filtering may help here. I wonder if it is possible to combine collaborative filtering and Markov Chain approach?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.