There's a lot of things happening on the chronolog. My actions trigger your reactions and this interaction can produce information. When I started thinking about statistics for the chronolog and bite into problem, I was very happy to find out I have quite a few possibilities. But because I always wanted to develop a recommendation engine, I got a bit carried away and first produced the Hot on the chronolog. Now the time has finelly come to go a little deeper and share insight to the flow of the chronolog.
The statistics following start with the basic overview of my activity, but as they progress, they become more and more complicated and interesting because they also include your feedback. Note that the statistics below are live, so perhaps some of my comments and assumptions are already wrong or obsolete when you are reading this.
The overview displays all the basic types of posts ever made on the chronolog - blog posts, bookmarks, tweets, loved music, rated movies and shouts. Boring stuff, but you have to start somewhere.
Now we add dynamics to the story, as time component is introduced. Trends analysis displays my activity of different types from month to month.
When I add you to the story, your interaction brings a whole new level of opportunities into the analysis. Here we can see the number of different actions (visits, likes, comments) for different post types, which is the base for calculating the importance of them. Because the stupid freeware chart doesn't support logarithmic scale, values on the chart have been square rooted for better comparison.
Here we give weight or ponders to specific interaction types (visits, likes, comments), based on their frequency, with a square root modifier. It is obvious that visits are the most frequent, because they need the simplest reaction (curiousity is enough). Likes follow, because they require an additional click. The comments are the rarest, because engagement is required, hence they have the biggest ponder and power. The data gathered in this step is used later on for the most complex analysis.
Based on the ponders above, we can calculate popularity of different categories (this goes only for blog posts). Average visits, likes and comments are modified with the number of posts in that category, so the results are more realistic. If you want to know how this algorithm works on individual posts on Hot on the chronolog, you can check out this post.
We finish off with the most complex and the coolest one, which introduces a bit of semantics and text mining to the analysis. All at least three letter words are competing, only 'the', 'and' and 'for' are left out. Based on the occurencies of these words in the titles, modified with ponders calculated above, we get an overview about which words are most appealing for the reader. This report may not be 100% perfect and real, but for now it works as a prototype for studying behaviour and dynamics.
Statistics and analytics are useful and fun, both in business environments and casual playgrounds. This is my first set, I hope I will go even further with these as the chronolog progresses. If you have any cool ideas on what else could be fun and insightful, let me know, I will be more than grateful to get a different view on the concept of the chronolog. Information must flow.