Estimating Derivatives of a Function at a Point

A derivative is a slope, and you can estimate a slope from any two points on a curve even without knowing the function's formula. Given a table of values, the average rate of change between two nearby x-values approximates the instantaneous rate of change at a point between them. This is just the slope formula, rise over run, applied to table entries instead of an equation.

The choice of which two points to use matters. A one-sided difference, like [f(a+h) - f(a)] / h, uses the point at a and a point on just one side of it, so it only captures how the function behaves moving in one direction. A symmetric difference, [f(a+h) - f(a-h)] / (2h), uses one point before a and one point after, straddling a equally. Because it balances information from both sides, it cancels out more of the curvature error and gives a closer estimate of the true slope at a, especially when the function bends noticeably between the given points.

This topic is the practical, data-driven counterpart to the formal limit definition of the derivative covered just before it. The difference quotient here is exactly the expression inside that limit, just evaluated at a small but nonzero h instead of pushed to zero. Later, when functions are given by formulas instead of tables, these same difference quotients reappear as the mechanism behind numerical derivative approximations and error analysis.