In producing a kernel curve, the smoothing parameter, which is related to bandwidth, controls the smoothness of the fit. Higher values (close to 100) produce curves that are almost straight lines, while lower values (close to 0) produce curves with a great deal of flexibility. It is often necessary to try several values before finding one that fits your data well.
MISE is the value of the smoothing parameter that minimizes the mean square error using generalized cross validation.
By default, the MISE is used as the smoothing parameter.