Kernel Density Estimation:

 KDE is the smoothing of random variable into a PDF.

It's called a non-parametric way of estimating PDF of RV. At each point kernels for that respective point add up to form the KDE curve.

The method of kernel density estimation revolves around the idea of selecting the optimal bandwidth for estimation. A very low bandwidth may result in a highly zagged tendency of the PDF & choosing a very high bandwidth may not capture the pattern of data properly.