Model Performance

Indices of model performance for linear and logistic regression models. The output depends on the form of the model (linear or logistic regression).

performance(x, ...)

Arguments

x

an object of type "lm" or "glm".

...

further arguments passed to or from other methods.

Value

The results of the methods performance.lm or performance.glm

Details

The performance function can be used to evaluate the predictive performance of a model with new data. If a data frame is not specified, the performance is evaluated on the training data (i.e., the data contained in the model component of the lm or glm object).

Examples

#######################
# multiple regression #
#######################
fit <- lm(mpg ~ hp + wt + accel + origin, data = auto_mpg)
performance(fit)
#> Multiple Regression Performance 
#> Data:  auto_mpg 
#> N:     388 
#> 
#> Model: lm(formula = mpg ~ hp + wt + accel + origin, data = auto_mpg) 
#> 
#>     RMSE Rsquared      MAE 
#>   4.1385   0.7184   3.1518 

#######################
# logistic regression #
#######################
fit2 <- glm(caesarian ~ age + bp + delivery.time, family = binomial, data = caesarian)
performance(fit2)
#> LOGISTIC REGRESSION PERFORMANCE 
#> Data              : caesarian 
#> N                 : 80 
#> Response variable : caesarian 
#> Category Balance  : no (0.42%) yes (0.58%) 
#> Predicted category: yes 
#> Prob to classify  : >=0.5
#> 
#> Model: glm(caesarian ~ age + bp + delivery.time,
#>            family = binomial, data = caesarian)
#> 
#> Confusion Matrix 
#> 
#>          Actual
#> Predicted no yes
#>       no  16   9
#>       yes 18  37
#> 
#> Overall Statistics 
#> 
#> Accuracy: 0.6625 
#> 97% CI  : (0.5481, 0.7645)
#> No Information Rate:  0.575 
#> P-Value [Acc > NIR]:  0.06961 
#> 
#> Statistics by Category 
#>                      
#> Sensitivity    0.8043
#> Specificity    0.4706
#> Pos Pred Value 0.6727
#> Neg Pred Value 0.6400
#> F1             0.7327
#> ---
#> Note: recall = sensitivity, 
#>       precision = pos pred value.