Exploring the Use of Variable Bandwidth Kernel Density Estimators

Abstract

Variable bandwidth kernel density estimators increase the window width at low densities and decrease it where data concentrate. This represents an improvement over the fixed bandwidth kernel density estimators. In this article, we explore the use of one implementation of a variable kernel estimator in conjunction with several rules and procedures for bandwidth selection applied to several real datasets. The considered examples permit us to state that when working with tens or a few hundreds of data observations, least-squares cross-validation bandwidth rarely produces useful estimates; with thousands of observations, this problem can be surpassed. Optimal bandwidth and biased cross-validation (BCV), in general, oversmooth multimodal densities. The Sheather-Jones plug-in rule produced bandwidths that behave slightly better in this respect. The Silverman test is considered as a very sophisticated and safe procedure to estimate the number of modes in univariate distributions; however, similar results could be obtained with the Sheather-Jones rule, but at a much lower computational cost. As expected, the variable bandwidth kernel density estimates showed fewer modes than those chosen by the Silverman test, especially those distributions in which multimodality was caused by several noisy minor modes. More research on the subject is needed.

Finite Element Analysis of the Proximal Femur and Hip Fracture Risk in Older Men

“…Whereas histograms are useful for discrete variables, smooth density functions better represent data from continuous distributions. KDE plots allowed us to examine the data for skewness and multimodality.Variables included in the best fitting models were then included in a logistic regression model (with weights). Analysis of the receiver operating characteristics (ROC) curve was performed to calculate area under the curve (AUC), and partial AUC values were also calculated because these are more sensitive to diagnostic classification in the more clinically relevant high specificity regimen., ROC curves were also calculated for unadjusted variables.…”

Increasingly certain about uncertainty: Intolerance of uncertainty across anxiety and depression

“…A Gaussian function was used with a bandwidth of 1 to compile the curves. Visual inspection of the plot revealed how distribution features (e.g., variance, skew, kurtosis) and modality (e.g., relative normality, bimodality) differed across diagnostic groups (Salgado-Ugarte, Shimizu, & Taniuchi, 1994).…”

Seroprevalence of HPV serotypes 6, 11, 16 and 18 in unvaccinated children from Mexico City

“…Thus in essence, we considered any results below LOD to be almost zero values, following Durazo and Sempos's argument [20]. Furthermore, the distribution of each of these antibody concentrations was estimated via kernel density estimators with the Epanechnikov kernel function, after randomly imputing values below the LOD [21]. …”