fit_linear_model_groups: Linear model fitting on subgroups of data

In sitscholl/rebecka_package: Utility functions used in the REBECKA project

Description

This function allows you to fit a (multiple) linear model between dependend and independend variables and perform cross validation to variable subgroups of the data.

Usage

fit_linear_model_groups(
  df,
  frml,
  group.cols = NULL,
  rowname.col = NULL,
  min.size = 30,
  output = "full",
  ...
)

Arguments

df	data.frame, A data.frame that is used to fit a linear model between dependent and independent variables.
frml	formula, A formula object for building the linear model
group.cols	string, column names that are used to group df
rowname.col	string, name of the column that is used to give rownames to the nested dataframes within df. Only useful if the desired output is 'augment' as the rownames are used to name the output rows.
min.size	integer, minimal amount of observations that have to be within a group in order to fit the linear model. If the group is smaller, it will be discarded.
output	string, one of 'full', 'minimal', 'augmented', 'glanced' or 'tidied'. 'full' will give all columns, 'minimal' only the linear model fits. For further information about the rest see the respective function in package::broom.
...	further arguments passed on to caret::trainControl (method, number, repeats). Defaults correspond to repeated crossvalidation using 5 folds repeated for 3 times.

Examples

fit_linear_model_groups(df = df.interp,
  frml = huglin ~ elevation,
  predictor.raster = dem.st,
  file.name = data/huglin.tif,
  set.zero = T)

sitscholl/rebecka_package documentation built on Aug. 25, 2020, 4:20 a.m.