Comparison of optimal bandwidths for some density estimators

Density estimation is a classic problem and has been extensively studied in Statistics. In this paper we briefly review some usual density estimators as the histogram, the Averaged Shifted Histograms (ASH) and the kernel density estimator (Kde). We care in particular on the choice of the bandwidth of Kde which is a fundamental parameter of the model and greatly influences the prediction. We finished our work with a simulation comparing the described methods on a common set of densities.