Tools for visualising densities, both discrete (binned histograms) and continuous (smooth density estimates).
All methods accept weights.
A discrete density is described by a tiling of the interval (1d) or plane (2d), along with a count of observations in each tile.
Binning data in 1d and 2d is tedious and tricky if you want to correctly deal with floating point (FP) issues. The `bin' package provides a fast and convenient interface for break calculation and binning in 1d and 2d.
bin_interval: interval bins (1d)bin_rect: rectangular bins (2d)bin_hex: hexagonal bins (2d)
A discrete density is described by a function that maps location on the real interval or plane to the density at that location.
This package also provides two methods for continuous density estimation, kernel_density and local_density. local_density uses local regression as implemented in the locfit package, and provides a large number of options to control the output. However, it can be slow, so kernel_density provides a faster implementation that offers less control.
kernel_density_1dkernel_density_2dlocal_density_1dlocal_density_2d
Both functions work with either 1d or 2d data, and both share a grid argument which specifies the locations where densities should be computed. This adds flexibility, allowing the function to be used to display the density over a regular grid, or just where the data points lie.