A constrained, non-linear fitting procedure was used, maximising the likelihood function of the logs of tumour volumes, with Gaussian distribution of data errors, taking their standard deviation as a parameter.
Using a Bayesian formulation of the problem enables relevant parameters to be estimated without recourse to maximising the likelihood function directly [4], which would be impractical for our application given the large number of raters and incomplete data structure.
The parameter estimates are then obtained by maximising the likelihood function usually using the Expectation-Maximization algorithm or its variations.
Furthermore, θ is a vector of eight parameters whose values we estimate by maximising the likelihood function, (7).
The effect of allowing annual variation in the base-line survival Ŝ 0 is then assessed by maximising the log-likelihood function (2) ln L = ∑ y ∑ i N s u r v, i ln Ŝ y, i + (N t o t a l, i - N s u r v, i ) ln (1 - Ŝ y, i ) Where Ŝ y, i = Ŝ 0, y e - B ^ F i and y = 1980, 1981...2009.
The feature weights and score threshold are selected to maximise the objective function.
The optimal solution maximises the objective function and minimises the weighted average constraints violations.
Two BSs form a cluster if the pair of them maximise the preference function.
The value of variance that maximises the likelihood function is 0.17 and 0.08.
In [8, 9], it is proposed to divide the time axis into decision windows and to maximise the objective function in each of these intervals, independently.
