Bind metric result data frames using rbindlist function

This function binds together data frames created from the create_metric_df() function using the rbindlist() function from the data.table package. The data frames are created by fitting linear models to the training data using the predictor variables provided in model_list, and then calculating the specified metric on the test data.

Usage

metric_bind(
  train_df,
  test_df,
  metric,
  method,
  kmin = "NA",
  target_variable,
  model_list
)

Arguments

train_df: A data frame containing the training data
test_df: A data frame containing the test data
metric: A character string indicating the metric to be calculated. Possible values are "rmse", "mae", "r2"
method: A character string indicating the type of model to be fitted. Possible values are "lm" (for linear regression) and "glm" (for generalized linear models).
kmin: An integer indicating the minimum value of k for k-fold cross-validation. If set to "NA", no cross-validation is performed.
target_variable: A character string indicating the name of the target variable.
model_list: A list containing the names of the predictor variables to be used in the linear models.

Value

A data frame containing the metric results for each predictor variable.

Examples

# load data
data(mtcars)

train_df <- mtcars[1:16, ]
test_df <- mtcars[17:32, ]

single_predictors <- list("mpg", "cyl", "disp", "hp", "am")
multiple_predictors <- list(c("mpg", "cyl"), c("disp", "am"), c("cyl", "am") )

# Single predictor lm regression model
metric_bind(train_df=train_df, test_df=test_df, metric="rmse", method="lm", kmin="NA", target_variable='gear', model_list=single_predictors)
#>    outcome predictor metric metric_value method kmin
#> 1:    gear       mpg   rmse    0.8595481     lm   NA
#> 2:    gear       cyl   rmse    0.8876402     lm   NA
#> 3:    gear      disp   rmse    0.8426984     lm   NA
#> 4:    gear        hp   rmse    1.1698191     lm   NA
#> 5:    gear        am   rmse    0.5899178     lm   NA

# Single predictor kknn regression model
metric_bind(train_df=train_df, test_df=test_df, metric="rmse", method="kknn", kmin=4, target_variable='gear', model_list=single_predictors)
#>    outcome predictor metric metric_value method kmin
#> 1:    gear       mpg   rmse    0.9291293   kknn    4
#> 2:    gear       cyl   rmse    0.8860904   kknn    4
#> 3:    gear      disp   rmse    0.8660254   kknn    4
#> 4:    gear        hp   rmse    1.0532687   kknn    4
#> 5:    gear        am   rmse    0.7126096   kknn    4
# Multiple predictors lm regression model
metric_bind(train_df=train_df, test_df=test_df, metric="rmse", method="lm", kmin="NA", target_variable='gear', model_list=multiple_predictors)
#>    outcome predictor metric metric_value method kmin
#> 1:    gear mpg + cyl   rmse    0.8839327     lm   NA
#> 2:    gear disp + am   rmse    0.7345654     lm   NA
#> 3:    gear  cyl + am   rmse    0.7427264     lm   NA

# Multiple predictors kknn regression model
metric_bind(train_df=train_df, test_df=test_df, metric="rmse", method="kknn", kmin=8, target_variable='gear', model_list=multiple_predictors)
#>    outcome predictor metric metric_value method kmin
#> 1:    gear mpg + cyl   rmse    0.9153637   kknn    8
#> 2:    gear disp + am   rmse    0.7673768   kknn    8
#> 3:    gear  cyl + am   rmse    0.8483670   kknn    8