kde2d_scikitlearn#

agabpylib.densityestimation.kde.kde2d_scikitlearn(xdata, ydata, Nx=100, Ny=100, xeval=None, yeval=None, xlims=None, ylims=None, evalOnData=False, kde_bandwidth=1.0, **kwargs)#

Provide a 2D kernel density estimate for a set of data points (x_i, y_i). Make use of the scikit-learn scikitlearn.neighbours.KernelDensity class.

Parameters:

  • xdata (float array) – 1D array of values of x_i

  • ydata (float array) – 1D array of values of y_i

  • xlims (tuple) – limits in x to use (xmin, xmax)

  • ylims (tuple) – Limits in y to use (ymin, ymax)

  • Nx (int) – Number of KDE samples in X (regular grid between xmin and xmax)

  • Ny (int) – Number of KDE samples in Y (regular grid between ymin and ymax)

  • xeval (float array) – Evaluate on this set of x coordinates (takes precedence over regular grid)

  • yeval (float array) – Evaluate on this set of y coordinates (takes precedence over regular grid)

  • evalOndata (boolean) – If true return the log(density) evaluated on the data (instead of the regular grid)

  • kde_bandwidth (float) – Bandwith for density estimator

  • **kwargs (dict) – Extra arguments for KernelDensity class initializer

Returns:

log_dens – The log(density) evaluated on the regular grid (shape (Nx,Ny)), or the log(density) evaluated for the data points (shape (xdata.size,)).

Return type:

float array