maksimiennustemallin
Maksimiennustemallin is a term used to describe a class of predictive models that estimate future outcomes by maximizing a likelihood-based objective. In this approach, a statistical model is specified for the data, typically denoted p(y|x, θ), where θ represents the model parameters. The goal is to choose θ so that the observed data are as likely as possible, i.e., to maximize the likelihood L(θ) = ∏ p(y_i|x_i, θ) or, equivalently, the log-likelihood ℓ(θ) = ∑ log p(y_i|x_i, θ).
Parameter estimation is usually achieved by optimization methods that find the maximizer of the (penalized) likelihood.
Maksimiennustemallin covers a wide range of model families, including generalized linear models, survival models, time-series models,
See also: likelihood, maximum likelihood estimation, probabilistic forecasting, information criteria.