Example of a Likelihood Function

Deterministic Model

We assume a deterministic model of this form:

\[model(x, \theta) = \theta_1 + \theta_2 x + \theta_3 x^2\]

Data

For this demonstration we uses simulated data:

Likelihood function

If we assume independent obserations, the likelihood of all observation is given by:

\[p(y \mid \theta, \sigma, x) = \prod_i p(y_i \mid \theta, \sigma, x_i)\]

For computational reasons we prefere to work with the log if the likelihood:

\[\log p(y \mid \theta, \sigma, x) = \log \prod_i p(y_i \mid \theta, \sigma, x_i) = \sum_i \log p(y_i \mid \theta, \sigma, x_i)\]

We can set the parameters and visualize the different components contributing to the likelihood function:

\[\hat{y} = \theta_1 + \theta_2 x + \theta_3 x^2\]