mn_log_loss: Mean log loss for multinomial data

In yardstick: Tidy Characterizations of Model Performance

Mean log loss for multinomial data

Description

Compute the logarithmic loss of a classification model.

Usage

mn_log_loss(data, ...)

## S3 method for class 'data.frame'
mn_log_loss(
  data,
  truth,
  ...,
  na_rm = TRUE,
  sum = FALSE,
  event_level = yardstick_event_level(),
  case_weights = NULL
)

mn_log_loss_vec(
  truth,
  estimate,
  na_rm = TRUE,
  sum = FALSE,
  event_level = yardstick_event_level(),
  case_weights = NULL,
  ...
)

Arguments

data	A data.frame containing the columns specified by truth and ....
...	A set of unquoted column names or one or more dplyr selector functions to choose which variables contain the class probabilities. If truth is binary, only 1 column should be selected, and it should correspond to the value of event_level. Otherwise, there should be as many columns as factor levels of truth and the ordering of the columns should be the same as the factor levels of truth.
truth	The column identifier for the true class results (that is a factor). This should be an unquoted column name although this argument is passed by expression and supports quasiquotation (you can unquote column names). For ⁠_vec()⁠ functions, a factor vector.
na_rm	A logical value indicating whether NA values should be stripped before the computation proceeds.
sum	A logical. Should the sum of the likelihood contributions be returned (instead of the mean value)?
event_level	A single string. Either "first" or "second" to specify which level of truth to consider as the "event". This argument is only applicable when estimator = "binary". The default uses an internal helper that defaults to "first".
case_weights	The optional column identifier for case weights. This should be an unquoted column name that evaluates to a numeric column in data. For ⁠_vec()⁠ functions, a numeric vector, hardhat::importance_weights(), or hardhat::frequency_weights().
estimate	If truth is binary, a numeric vector of class probabilities corresponding to the "relevant" class. Otherwise, a matrix with as many columns as factor levels of truth. It is assumed that these are in the same order as the levels of truth.

Details

Log loss is a measure of the performance of a classification model. A perfect model has a log loss of 0.

Compared with accuracy(), log loss takes into account the uncertainty in the prediction and gives a more detailed view into the actual performance. For example, given two input probabilities of .6 and .9 where both are classified as predicting a positive value, say, "Yes", the accuracy metric would interpret them as having the same value. If the true output is "Yes", log loss penalizes .6 because it is "less sure" of it's result compared to the probability of .9.

Value

A tibble with columns .metric, .estimator, and .estimate and 1 row of values.

For grouped data frames, the number of rows returned will be the same as the number of groups.

For mn_log_loss_vec(), a single numeric value (or NA).

Multiclass

Log loss has a known multiclass extension, and is simply the sum of the log loss values for each class prediction. Because of this, no averaging types are supported.

Author(s)

Max Kuhn

See Also

Other class probability metrics: average_precision(), brier_class(), classification_cost(), gain_capture(), pr_auc(), roc_auc(), roc_aunp(), roc_aunu()

Examples

# Two class
data("two_class_example")
mn_log_loss(two_class_example, truth, Class1)

# Multiclass
library(dplyr)
data(hpc_cv)

# You can use the col1:colN tidyselect syntax
hpc_cv %>%
  filter(Resample == "Fold01") %>%
  mn_log_loss(obs, VF:L)

# Groups are respected
hpc_cv %>%
  group_by(Resample) %>%
  mn_log_loss(obs, VF:L)


# Vector version
# Supply a matrix of class probabilities
fold1 <- hpc_cv %>%
  filter(Resample == "Fold01")

mn_log_loss_vec(
  truth = fold1$obs,
  matrix(
    c(fold1$VF, fold1$F, fold1$M, fold1$L),
    ncol = 4
  )
)

# Supply `...` with quasiquotation
prob_cols <- levels(two_class_example$truth)
mn_log_loss(two_class_example, truth, Class1)
mn_log_loss(two_class_example, truth, !!prob_cols[1])

yardstick documentation built on June 22, 2024, 7:07 p.m.