You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When working with the browser tracking data that Judith and colleages collected, for a while I considered relying primarily on concepts (and code libraries/tools) for network analysis, such as igraph for analysis. For various reasons we didn't really do this, with the result that our code feels somewhat home brew to me (in essence, we turned logs of visited sites into "grams" such as "google.com -> facebook.com" and then worked with that).
I would be interested in whether people at the workshop are considering seriously relying on network analysis for studying browser clickstream data and if not, what alternative strategies they have. What we did feels in retrospect like reinventing the wheel, and I personally feel better relying on well-developed packages for some of the very generic data processing issues involved, but on the other hand graph analysis has its own caveats. I know that Sandra Gonzales-Bailon and colleages have used such an approach with ComScore data, but I think in that particular case the approach fit well with their interest in the centrality of news sources.
In any case, I think a dedicated library for turning clickstream information in table format into something more meaningful along with some pretty plots would be useful. We wrote a few functions that point into this direction, but nothing comprehensive yet.
The text was updated successfully, but these errors were encountered:
Won't make it to the workshop because of family leave, but hope this can be
picked up on another occasion (ICA?). And yes, "statistical analysis of
clickstream data" sounds exactly right.
All the best,
Cornelius
On Tue, Oct 16, 2018 at 2:40 PM Wouter van Atteveldt < ***@***.***> wrote:
Great idea! I guess you are not coming to host this right? :-)
I think we could even broaden this a bit to "statistical analysis of
clickstream data" or something like that and pull in e.g. time series
analysis?
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#6 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/ALzPZx1kI4ZRQxTzJJIrq26FlWZI67KDks5uldO_gaJpZM4WpPE7>
.
When working with the browser tracking data that Judith and colleages collected, for a while I considered relying primarily on concepts (and code libraries/tools) for network analysis, such as igraph for analysis. For various reasons we didn't really do this, with the result that our code feels somewhat home brew to me (in essence, we turned logs of visited sites into "grams" such as "google.com -> facebook.com" and then worked with that).
I would be interested in whether people at the workshop are considering seriously relying on network analysis for studying browser clickstream data and if not, what alternative strategies they have. What we did feels in retrospect like reinventing the wheel, and I personally feel better relying on well-developed packages for some of the very generic data processing issues involved, but on the other hand graph analysis has its own caveats. I know that Sandra Gonzales-Bailon and colleages have used such an approach with ComScore data, but I think in that particular case the approach fit well with their interest in the centrality of news sources.
In any case, I think a dedicated library for turning clickstream information in table format into something more meaningful along with some pretty plots would be useful. We wrote a few functions that point into this direction, but nothing comprehensive yet.
The text was updated successfully, but these errors were encountered: