hot_stories.ipynb

Hi David, thanks for this. I think we are on the right path to get right metrics for this and the method you propose seems ok to me. A few questions and some issues of procedure to get this going this very week:

1. What is the period of time we are using? Do you collect and analyze the data once and it is like a snapshot or do you collect it throughout the week? If the latter, we would need to use time stamps within the day, as we want to draw the figures of each of the main categories of hotness that you have above and plot it to see the shapes of winners over time.
2. We are creating a total score score, by adding every week reads+comments+votes, and also some thing more nuanced like comments/reads and votes/reads. dago, pls, collect this and put it on a excel.
3. One thing that stands out of the last piece your analysis is how similar the numbers are through each category for "5SOS Preferences". Any ideas? What is the time period here?
4. Main issue with the method: if we do the social media collection on the Monday after they have been placed in the hot list, it is going to be wrong because the most we will get is what they do to "stay" in the list, but not to get there. My initial suggestion is that we do it daily, but we don't solve the problem, we just reduce the time error. Any ideas. Can we get more than the hot list, for instance the 20 top stories assuming that the hot list is made up if people who first goes through the top 20? If we keep this long enough, we could try to predict using ML the hot list members before they make the list.
   Method (Dago):
   Following Dave's proposed method:

1. at what time do we share the top 10 stories each Monday?
2. Within the hour, Dago locates the usernames of the authors across all social media and sends that back to Dave.
3. Dave, do we collect the week's social media activity with script, or manually?
4. Dago, in any case, create a excel, authors/weeks, and record there the most salient interventions they do in social media, to have a detailed look at that promotional content before we do the mass analysis.
5. Hopefully, the list will increase and we will have more than 10 authors after a few weeks, but the idea is that we keep track of all of them even if they drop from the list, so that we can control for ups and downs and the connection with social media activity.
6. Is a 10 week period starting this month ok?
   Any ideas, comments, suggestions?
   JL