Heat map of the kernel regression sensitivity analysis, depicting the absolute error in frequency estimation as a function of sample size (y-axis) and amplitude (x-axis). Lighter shades correspond to lower absolute errors, with cells marked by a red cross representing an average absolute error above 0.03, deemed as unreliable estimates. The heat map suggests that approximately ten observations are sufficient for a reliable frequency estimate across various amplitude levels, with the only exception being the lowest amplitude level.