Skip to contents

Fit a model to data

Source: R/fit_model.R
fit_model.Rd

A function to fit multiple feature of a model at the same time.

The feature of a model to be fit are passed as text strings in a list, called feature.

For some of these feature, fit_model can be configured to fit only some of the parameters or to lump some parameters together (e.g. IRS rounds that use the same pesticide). Each feature sets its own default for indexing the subset of parameters to be fit, but these can be overridden by passing text strings in options.

If length(feature) == 1 then there is another fit_ function in ramp.work that is equivalent to fit_model where parameter constraints and indexing options are described in greater detail. These are linked in the bullet points below.

If length(feature) > 1 then class(feature)= "multifit" and each feature in the list gets handled separately.

The function acts only on the following text strings in feature:

  • mean_forcing : fits the mean forcing parameter (also, see fit_mean_forcing)

  • trend : fits interpolation points for a spline function (also, see fit_trend)

    • by default, all \(y\) values of interpolation points are fitted

    • a subset can be configured by setting trend_ix in options

    • if trend and mean_forcing are both called, at least one of the interpolation points should be set to 1

  • At most one of the following aspects of seasonality:

    • phase : the time of year when forcing peaks (also, see fit_season_phase)

    • pw : a shape parameter affecting the amplitude (also, see fit_season_pw)

    • bottom : a shape parameter affecting amplitude (also, see fit_season_bottom)

    • amplitude : fits the shape parameters that affect amplitude: pw & bottom (also, see fit_season_amplitude)

    • season : fits all seasonality shape parameters: pw & bottom & phase (also, see fit_season)

  • irs_coverage : fit irs coverage parameters (also see fit_irs_coverage)

    • by default, coverage levels for all rounds are fitted independently

    • other options can be fitted by setting irs_ix in options

  • bednet_contact : fit bed net coverage parameters (also see fit_bednet_contact)

    • by default, coverage levels for all rounds are fitted independently

    • other options can be fitted by setting bednet_ix in options

The xds model object that gets returned by fit_model has the optimal paramters.

Usage

fit_model(xds_obj, feature, options = list())

Arguments

xds_obj

an xds model object

feature

a list of one or more model feature to fit

options

a list of options to override feature-specific defaults

Value

an xds model object