Assume that pdf/pmf for the random variable belongs to the exponential family. Then will the random variable under some one-to-one transformation also belong to the exponential family? To assess this, recall that when transforming a random variable, the rate of change between and needs to be accounted for via the Jacobian when deriving the pdf/pmf for :
Substituting the exponential family pdf/pmf for and the Jacobian, , gives:
Here, does not depend on and is in its own right a pdf/pmf as it is a transformation of the pdf/pmf . Furthermore, since the transform also does not depend on , the pdf/pmf also belongs to the exponential family. The function is called the sufficient statistic for the parameter .
A statistic is a sufficient statistic for the parameter if the conditional distribution of the data given the statistic does not depend on the parameter , i.e.:
Exercise 3.25
The pdf for the random variable is:
Show that this belongs to the exponential family. What is the sufficient statistic for the canonical parameter?