Estimation of Mode on the Basis of a Truncated Sample
MetadataShow full item record
The problem of estimation of the mode of a continuous distribution (unction is considered. The estimation of the mode based on estimators of the density function is well known, see e.g. Härdle (1991) and Parzen (1962). New parameters of the continuous distribution function will be defined: the quasi-mode and mean median. They are parameters of the appropriately truncated random variable. Next, the estimators of the mode, such as the sample quasi-mode or sample mean-median, are determined. These statistics are usually biased estimators of the mode. Well known “jackknife” procedure is adapted to estimate Their mean square error. The accuracy of the mode estimation is studied on the basis of computer simulation.