I have user data of a website which I like to analyse with data mining techniques. But before I can do than, I have to transform the data.
The data contains a visit id, timestamps for a specific action like a click and the url on which the click took place, and other descripitive data.
An example file is attached.
Is there a possibility in knime to
a) reconstruct this user flow and give a good visualisation
b) make calculations to receive metric data. (i.e. count(clicks), avg. time per site, time per visit, etc.)
I could use sql, but since every flow is of different length, it is quite a hazard.
Thank you in advance!
Good morning Andra,
Of course there are possibilities in KNIME for both of your goals, and of course there are many different approaches. As I am not able to see your example file I can only give you some ideas of how to approach this project:
- For visualization you could either use the network mining plugin, with your pages being the nodes and edges being a click on a link, or you could use the "Parallel Coordinates"-Node, with your sides on the y- and the steps in the flows on the x-axis.
- For the general metrics the "Group By"-Node is the tool you want to use! Selecting the id of the page as the Group column you can set the aggregation such that it calculates whatever you like. This example calculates the click cout per page and the average time spent on this page.
Again, these are just some ideas to get your project started. If you have more questions don't hesitate to ask.