valloss: Weights estimation using different model combinations...

In kessysalvatory/CoMoMo: CoMoMo combines multiple mortality forecasts using different model combinations

Weights estimation using different model combinations methods.

Description

We consider four different model combination approaches: Simple Model Averaging, Bayesian Model Averaging, Model Confidence Set, and Stacked Regression Ensemble. These methods vary from each other depending on how they use the historical data to choose the combining weights or the models to be combined.

Usage

valloss(
  models,
  data = NULL,
  Dxt = NULL,
  Ext = NULL,
  ages.fit = NULL,
  years.fit = NULL,
  ages = NULL,
  years = NULL,
  holdout = NULL,
  h = NULL
)

Arguments

models	are the specified models to be combined.
data	an optional object of type StMoMoData containing information on deaths and exposures to be used for training the model. This is typically created with function StMoMoData. If this is not provided then the training data is taken from arguments, Dxt, Ext, ages, years.
Dxt	optional matrix of deaths data.
Ext	optional matrix of observed exposures of the same dimension of Dxt.
ages.fit	optional vector of ages to include in the training. Must be a subset of ages.
years.fit	optional vector of years to include in the training. Must be a subset of years.
ages	optional vector of ages corresponding to rows of Dxt and Ext.
years	optional vector of years corresponding to rows of Dxt and Ext.
h	number of years for forecasting horizon.

Details

Simple Model Averaging assigns equal weights to all the models.

Bayesian Model Averaging estimates the weights using the posterior model probabilities.

Model Confidence Set chooses the subset of superior mortality models to combine where each model is assigned equal weight

Stacked Regression Ensemble combines point forecasts from multiple base learners using the weights that optimize a cross-validation criterion.

kessysalvatory/CoMoMo documentation built on Sept. 15, 2022, 8:12 a.m.