I would also want to filter out plane crashes caused by interference: suicides, bombings, hijackings, birdstrike and the like. This is because I fly happier knowing that mechanical or electrical error is so very very low. This would include however cases such as Air France Flight 447, which was a combination of faulty readings and human error. So, the recent Ethiopia and Lion air crashes would definitely be included, though maybe these were avoidable if the pilots were better educated? So I guess I'd need to see three categories: outside interference, pilot error, mechanical/electrical error.

The average person makes far more trips by car than by airplane.

Feel free to tear into this. I know it's flawed. But FWIW here's some back-of-the-napkin math I put together when discussing in an e-mail thread. Note however that it's slightly conservative WRT road fatalities. You can choose not to drive drunk, drive defensively, etc, and substantially reduce your risk because a large fraction of deaths are the drunk and other reckless drivers themselves.

  > > > > > In 2017 there were 0.06 deaths per million flights of large,
  > > > > > commercial airliners:
  > > > > > 
  > > > > >   https://www.reuters.com/article/us-aviation-safety/2017-safest-year-on-record-for-commercial-passenger-air-travel-groups-idUSKBN1EQ17L
  > > > > > 
  > > > > > By contrast, in 2009 white people averaged 1,500 trips per year:
  > > > > > 
  > > > > >   https://www.fhwa.dot.gov/policy/2010cpr/chap1.cfm
  > > > > > 
  > > > > > Multiply 1,500 trips/person by 360 million people (2009 U.S.
  > > > > > population), divide ~35,000 road fatalities by the number of trips,
  > > > > > and multiple by 1,000,000, and you literally get 0.06 fatalities per
  > > > > > million trips. Even if the numbers are slightly inaccurate
  > > > > > (different years, different demographics, wrong cohorts), the point
  > > > > > is that strictly speaking flying really isn't that much safer than
  > > > > > driving. Certainly nothing like the several orders of magnitude
  > > > > > claimed; not when using the more meaningful metric of fatalities per
  > > > > > trip.
  > ...
  > > > > I don't think that normalizes for the number of people on the plane,
  > > > > though.
  > > >
  > > > Hmmmmm. Good point... I think.
  > > >
  > > like, does a flight of 100 people count as a single "trip", or as 100? If
  > > it only counts as one, then you should normalize, because I'm sure they
  > > don't count each crash as a single fatality.
  > 
  > Here are numbers from 2018 from the original source:
  > 
  >   https://news.aviation-safety.net/2019/01/01/aviation-safety-network-releases-2018-airliner-accident-statistics/
  >   https://docs.google.com/spreadsheets/d/1SDp7p1y6m7N5xD5_fpOkYOrJvd68V7iy6etXy2cetb8/edit#gid=1448957446
  > 
  > That's 0.39 fatal *accidents* per million flights, based on 15 fatal
  > flights and 38,100,0000 departures. There were 556 total deaths from those
  > 15 fatal flights.[1]
  > 
  > The question, is how many passengers on average per flight. If we divide
  > the number of deaths (556) by the number of fatal flights (15), we get 37
  > on average. If we divide the number of total passengers for 2018 (4.3
  > billion[2]) by the total number of flights (38,100,000), we get 112
  > passengers on average.
  > 
  > So average number of fatalities per departure could be somewhere between
  > 0.01 and 0.003. Definitely lower than 0.06 but nowhere close to 750x
  > lower.
  > 
  > [1] I'm using passenger+cargo because there's only one figure given for
  > total number of departures and I assume that's the aggregate. Counting
  > just passengers we get 0.36 instead of 0.39 which isn't much of a
  > difference, probably because cargo flights have only 2 people on board.
  > Still, it would be nice to know the total number of passenger-only
  > departures.
  >
  > [2] https://www.statista.com/statistics/564717/airline-industry-passenger-traffic-globally/