Get predictions to asses model performance from a fitted plus model.

Get predictions to asses model performance from a fitted plus model.

Usage

get_predictions(object, newx, newy, plus = TRUE)

Arguments

object

A plus object with glmnet classification

newx

New data to predict the class

newy

The observed class to meassure the performance of the model

plus

Logical. If it is true use the cutoff threshold from the plus model. This tends to reduce the false negatives and increase the recall. Othewise gets the predicted value from glmnet default predicted response.#'

Value

A data.frame containing four columns. The truth independent value (observed y), the probability to belong in Class1, the probability to belong Class2 and the predicted class either the plus (use cutoff threshold) or glmnet model.

Examples

data(binexample)
x = binexample$x
y = binexample$y
s <- sample(seq(nrow(x)), 65, replace = FALSE)
train_x <- x[s, ]
train_y <- y[s]
newx <- x[-s, ]
newy <- y[-s]
model <- plus(train_x, train_y)
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
#> Warning: one multinomial or binomial class has fewer than 8  observations; dangerous ground
get_predictions(model, newx, newy)
#> # A tibble: 35 × 4
#>    truth   Class1  Class2 predicted
#>    <fct>    <dbl>   <dbl> <fct>    
#>  1 Class1 0.930   0.0703  Class2   
#>  2 Class2 0.778   0.222   Class2   
#>  3 Class2 0.989   0.0111  Class2   
#>  4 Class2 0.751   0.249   Class2   
#>  5 Class1 0.986   0.0138  Class2   
#>  6 Class1 0.998   0.00183 Class2   
#>  7 Class2 0.867   0.133   Class2   
#>  8 Class2 0.00335 0.997   Class2   
#>  9 Class1 0.676   0.324   Class2   
#> 10 Class1 0.996   0.00431 Class2   
#> # ℹ 25 more rows
get_predictions(model, newx, newy, plus = FALSE)
#> # A tibble: 35 × 4
#>    truth  Class1 Class2 predicted
#>    <fct>   <dbl>  <dbl> <fct>    
#>  1 Class1  0.879 0.121  Class1   
#>  2 Class2  0.838 0.162  Class1   
#>  3 Class2  0.906 0.0944 Class1   
#>  4 Class2  0.773 0.227  Class1   
#>  5 Class1  0.925 0.0754 Class1   
#>  6 Class1  0.958 0.0420 Class1   
#>  7 Class2  0.836 0.164  Class1   
#>  8 Class2  0.201 0.799  Class2   
#>  9 Class1  0.761 0.239  Class1   
#> 10 Class1  0.955 0.0448 Class1   
#> # ℹ 25 more rows