...

/

Maximum Likelihood Estimation and Its Applications

Maximum Likelihood Estimation and Its Applications

Get an introduction to maximum likelihood estimation and see examples using an interactive code in Python.

What is maximum likelihood estimation?

Maximum likelihood estimation (MLE) is a statistical technique for estimating the parameters of a given model by maximizing the likelihood so that the observed data is most probable. The basic idea behind MLE is to find the parameter values that make the observed data most likely. In other words, the goal is to find the parameter values that maximize the likelihood of observing the data. To do this, the model is first specified, then the likelihood of the data given to the model is computed. The parameters of the model are then adjusted to maximize the likelihood. The parameter values that maximize the likelihood are then taken as the estimates of the model parameters.

Assumptions

The main assumption required to consider before using MLE is that the data is i.i.d., that is, it is independent and identically distributed:

  • Identically distributed means that there are no overall trends—the distribution doesn’t fluctuate and all items in the sample are taken from the same probability distribution.

  • Independent means that the sample items are all independent events. In other words, they are not connected to each other in any way; knowledge of the value of one variable gives no information about the value of the other and vice versa.

Parametric and nonparametric MLE

There are two types of MLE, parametric and nonparametric. The comparison between the two types is given in the table below:

Comparison between Parametric and Nonparametric MLE

Property

Parametric MLE

Nonparametric MLE

Assumption

Data is generated from an underlying parametric model.

No underlying parametric model.

Specification of data

Finite set of parameters specifies the model estimated by MLE.

Estimation of PDF from the data.

Methods

Linear regression, logistic regression, and Poisson regression.

Kernel density estimation, nearest neighbor estimation, and k-means clustering.

  • Parametric MLE: It assumes that the data is generated from an underlying parametric model. This means that the model is specified by a finite set of parameters and these parameters can be estimated using MLE. The most commonly used parametric models are linear regression, logistic regression, and Poisson regression.

  • Nonparametric MLE: It does not assume an underlying parametric model. Instead, the model is estimated directly from the data. Nonparametric MLE is often used when the underlying data-generating process is unknown or too complex to ...