Clickstream analytics with Markov Chains - London Measurecamp #12

19/3/2018

This talk provides an overview of clickstream analysis and presents an example of using Markov chains to model the clickstream and provide transition probabilities from any page to any other. It also shows how to predict the user's next click given the sequence of previous page requests.

Clickstream measure camp12 from Alexandros Papageorgiou

2 Comments

Roshan link

20/3/2018 01:53:28 am

Thank you for this session. It's given me great insight into how attribution works. Do you think Regression can be combined with markov chains for mor efficient path analysis?

Alexandros Papageorgiou

20/3/2018 04:22:34 am

I am glad the session was useful to you. You can use regression to predict the absorbing state, purchase or exit,based on the sequence of previous pages visited. The clickstream package has a function called frequencies that can convert your clickstream into a compatible data structure for that type of regression. That said I wouldn't consider regression as the most natural approach to working with clickstreams- at least not to the same degree as other methods like network analysis or association rules for example

ALEXANDROS PAPAGEORGIOU

Clickstream analytics with Markov Chains - London Measurecamp #12

Leave a Reply.