Evaluation Metrics

Within the package, three primary evaluation metrics are implemented to assess the performance of fitted temporal mixture models: Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and Log-Likelihood. These metrics provide insights into the goodness-of-fit of the models while accounting for model complexity.

TemporalMixtureModels.loglikelihood — Method

loglikelihood(result::MixtureResult, t::AbstractVector{T}, y::AbstractMatrix{Y}, ids::AbstractVector{Int}; inputs=nothing) where {T<:Real, Y<:Union{Real, Missing}}

Compute the total log-likelihood of the fitted mixture model on the given data.

Arguments

result::MixtureResult: The result of fitting the mixture model.
t::AbstractVector: Time points of the observations.
y::AbstractMatrix: Observed data (vector or matrix).
ids::AbstractVector{Int}: Cluster assignments for each observation.
inputs: Optional additional inputs for prediction.

Returns

Float64: The total log-likelihood of the model on the data.

source

TemporalMixtureModels.aic — Function

aic(result::MixtureResult, t::AbstractVector{T}, y::AbstractMatrix{Y}, ids::AbstractVector{Int}; inputs=nothing) where {T<:Real, Y<:Union{Real, Missing}}

Compute the Akaike information criterion of the fitted mixture model on the given data.

Arguments

result::MixtureResult: The result of fitting the mixture model.
t::AbstractVector: Time points of the observations.
y::AbstractMatrix: Observed data (vector or matrix).
ids::AbstractVector{Int}: Cluster assignments for each observation.
inputs: Optional additional inputs for prediction.

Returns

Float64: The Akaike information criterion of the model on the data.

source

TemporalMixtureModels.bic — Function

bic(result::MixtureResult, t::AbstractVector{T}, y::AbstractMatrix{Y}, ids::AbstractVector{Int}; inputs=nothing) where {T<:Real, Y<:Union{Real, Missing}}

Compute the Bayesian information criterion of the fitted mixture model on the given data.

Arguments

result::MixtureResult: The result of fitting the mixture model.
t::AbstractVector: Time points of the observations.
y::AbstractMatrix: Observed data (vector or matrix).
ids::AbstractVector{Int}: Cluster assignments for each observation.
inputs: Optional additional inputs for prediction.

Returns

Float64: The Bayesian information criterion of the model on the data.

source