Multivariate Adaptive Regression Splines - Hinge Functions

Hinge Functions

Hinge functions are a key part of MARS models. A hinge function takes the form

or

where is a constant, called the knot. The figure on the right shows a mirrored pair of hinge functions with a knot at 3.1.

A hinge function is zero for part of its range, so can be used to partition the data into disjoint regions, each of which can be treated independently. Thus for example a mirrored pair of hinge functions in the expression


6.1 \max(0, x - 13)
- 3.1 \max(0, 13 - x)

creates the piecewise linear graph shown for the simple MARS model in the previous section.

One might assume that only piecewise linear functions can be formed from hinge functions, but hinge functions can be multiplied together to form non-linear functions.

Hinge functions are also called hockey stick functions. Instead of the notation used in this article, hinge functions are often represented by where means take the positive part.

Read more about this topic:  Multivariate Adaptive Regression Splines

Famous quotes containing the words hinge and/or functions:

    It is one of the consolations of philosophy that the benefit of showing how to dispense with a concept does not hinge on dispensing with it.
    Willard Van Orman Quine (b. 1908)

    One of the most highly valued functions of used parents these days is to be the villains of their children’s lives, the people the child blames for any shortcomings or disappointments. But if your identity comes from your parents’ failings, then you remain forever a member of the child generation, stuck and unable to move on to an adulthood in which you identify yourself in terms of what you do, not what has been done to you.
    Frank Pittman (20th century)