7:00 – 8:00 Research
- This Is Why Trump’s Conspiracy Theories Work, Say Experts
- Good advice on writing papers
- Setting up to get clustering data.
- Explore/Exploit ratio 50/50
- Explore will be in the flocking range: 0.1 – 1.6
- Exploit will be in the echo chamber stage: 3.2 – 10.0
- Cluster EPS of 0.25 gives good diversity
- Create a DataFrame from the “Cluster ID” sheet
- Ran the sims, but the default destination is wrong. Re-running
- Discriminate between explorer and exploiter cluster membership over time
- Clusters the agent belonged to
- Will need to post-process to fix cluster switching. Probably taking all the cluster average positions and applying the a common ID if the difference between the samples is less than a given delta
8:30 – 3:30 BRC
- Generating data for clustering
- Will try to read in and merge the DataFrames so that I have position (angle or origin distance) to calculate group persistence
- Human Motion Recognition Using Isomap and Dynamic Time Warping
- Fixing DataFrames. Mostly this is a case of bad data handling. This is in the data:
359000.00,836,2,2,4,0,1,0,2,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,#<Geocoder::Result::Bing:0x007fe9a1b39718>,#<Geocoder::Result::Bing:0x007fe9a1b39150>
Which means that we have to handle ‘#<Geocoder::Result::Bing:0x007fe9a1b39718>,#<Geocoder::Result::Bing:0x007fe9a1b39150>. I am very disappointed in this book
- Giving up. Wrote a review. Monday I’ll try doing joins on Dataframes. That being said, I learned a lot on how to check for errors.
‘