rcvirtual.strategy-class: Statistical model-fitting algorithms (strategy)

In rtlemos/rcvirtual: Virtual references classes for statistical modeling

Description

This is a virtual Reference Class for strategy RCs.

Details

This RC contains fields (a.k.a. "attributes") and methods (a.k.a. "procedures") to fit statistical models.

Fields

parameters

list. Parameters used by strategy.

model.res

list. Model fitting output

model.fitted

logical. Has the model been fitted?

Methods

construct()

Construct basic objects

get.bounds(param.name = "latitude")

Provides the bounds for a given parameter

get.data(param.name, field.name = "value")

Retrieve parameter data

get.dependencies(x)

Loads the dependencies of parameter x onto the global environment. Use this to debug 'get.compute.x' functions

get.imputation(Yall, f, Q, u)

Impute missing values from obs & forecasts. Yall: vector with some NAs (missing values); f: vector of predictive means; Q: matrix of predictive covariance; u: vector of Gaussian variates; output: list with indices of NAs and imputed values

get.log.likelihood()

Returns the log likelihood for a parameter vector

get.parameter.names()

Retrieve the names of parameters in the model, ordered according to the graph

get.parameter.types()

Retrieve the types (classes) of all model parameters, ordered according to the graph

get.time.formatted(tbounds.char, tstep = "day", tz = "GMT")

Function that formats time information: instants, days, months, years, dates, starting instant and ending instant.

get.value(param.name)

Shortcut function to retrieve a parameter value

is.model.fitted()

Check if the model has been fitted

is.parameter(param.name)

Check if param.name is among the list of model parameters

is.redundant.calculation(param.name = NA, parent.names = NULL)

Should a derived quantity be recomputed?

set.data(param.name, field, obj, update.timestamp = TRUE)

Store an object in a parameter

set.graph()

Generates a sorted Directed Acyclic Graph of model parameters

set.optimize(mle = TRUE, maxit = 1000)

Compute maximum likelihood estimates (MLE) or maximum a posteriori estimates (MAP) of unknown model parameters

set.parameters()

Constructs a list of parameters required to fit a model

set.value(param.name, obj, update.timestamp = TRUE)

Shortcut function to store an object in a parameter

rtlemos/rcvirtual documentation built on May 28, 2019, 9:56 a.m.