The log-likelihood is, as the term suggests, the natural logarithm of the likelihood.
To define the likelihood we need two things:
some observed data (a sample), which we denote by (the Greek letter xi);
a set of probability distributions that could have generated the data; each distribution is identified by a parameter (the Greek letter theta).
Roughly speaking, the likelihood is a function that gives us the probability of observing the sample when the data is extracted from the probability distribution with parameter .
We will provide below a rigorous definition of log-likelihood, but it is probably a good idea to start with an example.
In this case, the sample is a vectorwhose entries are draws from a normal distribution.
The probability density function of a draw iswhere and are the parameters (mean and variance) of the normal distribution.
The parameter vector is
The set of distributions that could have generated the sample is assumed to be the set of all normal distributions (that can be obtained by varying the parameters and ).
In order to stress the fact that the probability density depends on the two parameters, we write
The joint probability density of the sample is because the joint density of a set of independent variables is equal to the product of their marginal densities (see the lecture on Independent random variables).
The likelihood function is
In other words, when we deal with continuous distributions such as the normal distribution, the likelihood function is equal to the joint density of the sample. We will explain below how things change in the case of discrete distributions.
The log-likelihood function is
The log-likelihood function is typically used to derive the maximum likelihood estimator of the parameter .
The estimator is obtained by solvingthat is, by finding the parameter that maximizes the log-likelihood of the observed sample .
This is the same as maximizing the likelihood function because the natural logarithm is a strictly increasing function.
One may wonder why the log of the likelihood function is taken. There are several good reasons.
To understand them, suppose that the sample is made up of independent observations (as in the example above).
Then, the logarithm transforms a product of densities into a sum. This is very convenient because:
the asymptotic properties of sums are easier to analyze (one can apply Laws of Large Numbers and Central Limit Theorems to these sums; see the proofs of consistency and asymptotic normality of the maximum likelihood estimator);
products are not numerically stable: they tend to converge quickly to zero or to infinity, depending on whether the densities of the single observations are on average less than or greater than 1; sums are instead more stable from a numerical standpoint; this is important because the maximum likelihood problem is often solved numerically on computers where limited machine precision does not allow us to distinguish a very small number from zero and a very large number from infinity.
We finally give a rigorous definition of log-likelihood
The following elements are needed to define the log-likelihood function:
the distribution of belongs to a parametric family: there is a set of real vectors (called the parameter space) whose elements (called parameters) are put into correspondence with the distributions that could have generated ; in particular:
when the joint probability mass (or density) function is considered as a function of for fixed (i.e., for the sample we have observed), it is called likelihood (or likelihood function) and it is denoted by . So,if is discrete and if is continuous.
Given all these elements, the log-likelihood function is the function defined by
You will often hear the term "negative log-likelihood". It is just the log-likelihood function with a minus sign in front of it:
It is frequently used because computer optimization algorithms are often written as minimization algorithms.
As a consequence, the maximization problemis equivalently written in terms of the negative log-likelihood asbefore being solved numerically on computers.
More examples of how to derive log-likelihood functions can be found in the lectures on:
The log-likelihood and its properties are discussed in a more detailed manner in the lecture on maximum likelihood estimation.
Previous entry: Joint probability mass function
Next entry: Loss function
Please cite as:
Taboga, Marco (2021). "Log-likelihood", Lectures on probability theory and mathematical statistics. Kindle Direct Publishing. Online appendix. https://www.statlect.com/glossary/log-likelihood.
Most of the learning materials found on this website are now available in a traditional textbook format.