Articles • logmu

Get started
Analyse experience data
Fit a model: Fit a proportional hazards mortality model in R with logmu: covariates on a reference mortality, maximum likelihood fitting and weighting by amounts.
Test multiple models: Compare candidate mortality models with compare_models() in logmu, which ranks them by penalised log-likelihood (AIC), allowing for weights and overdispersion.
Cluster categories

Measures matter: Mortality experience analysis fundamentals, part 1: experience data, actual and expected deaths as measures, and why overdispersion must be allowed for.
A over E: Mortality experience analysis fundamentals, part 2: why A/E works, its expected value and variance, and A/E diagnostics.
Log-likelihood: Mortality experience analysis fundamentals, part 3: log-likelihood for mortality experience data, defined directly in terms of A and E.
Proportional hazards: Mortality experience analysis fundamentals, part 4: why proportional hazards models are probably all you need for mortality modelling.
Suddenly AIC: Mortality experience analysis fundamentals, part 5: choosing between mortality models using Akaike’s information criterion (AIC).
Overdispersion and quasi-log-likelihood: Mortality experience analysis fundamentals, part 6: the causes of overdispersion and how to allow for it using quasi-log-likelihood.
Incoherent rating factors: Mortality experience analysis fundamentals, part 7: why rating factors must be coherent, the four conditions for coherence, and pension as a rating factor.