Generally, experimental projects will collect noisy measured values ...
call them tabular values. Collect plenty to be representative.
This document is a guide line to projects that produce clean discrete.
For clean D3 data set must have 6 decimals at least.
1. F.B.Hildebrand is kind of 'yardstic' ... D3 has dropped from 101/78.
2. Discrete convolution is unscaled ... NOT offered.
3. Ridder agrees closely with Hlidebrand.
If you have hyper noisy data, low populated [Corn Thermies]
smooth wisely ... possibly fit, populate ... suite.

Maths Derivative Ridder DEMO Copy.sm (132 KiB) downloaded 76 time(s).

... augmented for convenience.

Maths Derivative Ridder DEMO Copy.sm (171 KiB) downloaded 75 time(s).