3 ms·

If you're willing to share: have you discovered any unexpected causal links yet?

• mg 5y ago
  Hard to say. My tool so far does not output p values. At the moment, it visualizes possible cause/effect relationships in kind of a dot cloud. It shifts all effects in time relative to the possible cause. So an effect would look like a dent in the graph on the right side of the origin. Some cause/effect pairs look interesting. But I want to gather more data and implement appropriate statistical tests before I publish something. It is an interesting question, which statistical tests to apply to this kind of open log format. If anybody here has ideas about this, I would be very interested in hearing them.
  • Blahah 5y ago
    Super interesting question! Some combination of NLP and classical time series stats will probably show lots of interesting things. What's the density over time of the entries? I.e. How many entries per day? And are you consistently recording the same kinds of events and information about them? How many different kinds of events?
    • mg 5y ago
      Have you seen the link I posted? I think it answers most questions and is a good starting point for a discussion.