Kernel estimation of a univariate probability density and of its derivatives of any order r, with data-driven bandwidth selection. Provides the kernel estimator of the derivative of order r, the kernels, their derivatives and convolutions, and eight bandwidth selectors: the asymptotic MISE rule (AMISE), and unbiased (UCV), biased (BCV1, BCV2), complete (CCV), modified (MCV), trimmed (TCV), smoothed (SCV) and maximum-likelihood (MLCV) cross-validation. Supports ten named kernels (Gaussian, Epanechnikov, uniform, triangular, biweight, triweight, tricube, cosine, Silverman, logistic) and the power family Pow(s,t), c(1-|x|^s)^t on [-1,1]. Results are S4 objects with ggplot2 graphics; the O(n^2) criteria are computed in C++ ('Rcpp') with OpenMP parallelism. Includes five example samples from Marron and Wand (1992) normal mixtures, a vignette and demos. Companion to Guidoum (2020) <doi:10.48550/arXiv.2012.06102>.