e.g. this occurs in the satellites predictive notebook (https://probcomp-1.csail.mit.edu/6d03d3f3cf4c0e22755f471f5ebfe5f6/20170627-oreilly/satellites-predictive.html) with the following query:
%bql .density --xmin=-20000 --xmax=60000 SIMULATE perigee_km FROM satellites LIMIT 250;
where the distribution of simulations seems trimodal, but the density curve looks distinctly bimodal.
Perhaps there is a UI change, such as accepting a different smoothing function as a parameter to .density, that could help resolve such quirks.
e.g. this occurs in the satellites predictive notebook (https://probcomp-1.csail.mit.edu/6d03d3f3cf4c0e22755f471f5ebfe5f6/20170627-oreilly/satellites-predictive.html) with the following query:
where the distribution of simulations seems trimodal, but the density curve looks distinctly bimodal.
Perhaps there is a UI change, such as accepting a different smoothing function as a parameter to
.density, that could help resolve such quirks.