library(bayesrules)
For Bayesian model evaluation, the bayesrules package has three functions prediction_summary()
, classification_summary()
and naive_classification_summary()
as well as their crossvalidation counterparts prediction_summary_cv()
, classification_summary_cv()
, and naive_classification_summary_cv()
respectively.
Functions  Response  Model 

prediction_summary() prediction_summary_cv()

Quantitative  rstanreg 
classification_summary() classification_summary_cv()

Binary  rstanreg 
naive_classification_summary() naive_classification_summary_cv()

Categorical  naiveBayes 
Given a set of observed data including a quantitative response variable y and an rstanreg model of y, prediction_summary()
returns 4 measures of the posterior prediction quality.
Median absolute prediction error (mae) measures the typical difference between the observed y values and their posterior predictive medians (stable = TRUE) or means (stable = FALSE).
Scaled mae (mae_scaled) measures the typical number of absolute deviations (stable = TRUE) or standard deviations (stable = FALSE) that observed y values fall from their predictive medians (stable = TRUE) or means (stable = FALSE).
and 4. within_50 and within_90 report the proportion of observed y values that fall within their posterior prediction intervals, the probability levels of which are set by the user. Although 50% and 90% are the defaults for the posterior prediction intervals, these probability levels can be changed with prob_inner
and prob_outer
arguments. The example below shows the 60% and 80% posterior prediction intervals.
# Data generation
< data.frame(x = sample(1:100, 20))
example_data $y < example_data$x*3 + rnorm(20, 0, 5)
example_data
# rstanreg model
< rstanarm::stan_glm(y ~ x, data = example_data, refresh = FALSE)
example_model
# Prediction Summary
prediction_summary(example_model, example_data,
prob_inner = 0.6, prob_outer = 0.80,
stable = TRUE)
mae mae_scaled within_60 within_801 3.540663 0.9855435 0.65 0.9
Similarly, prediction_summary_cv()
returns the 4 crossvalidated measures of a model’s posterior prediction quality for each fold as well as a pooled result. The k
argument represents the number of folds to use for crossvalidation.
prediction_summary_cv(model = example_model, data = example_data,
k = 2, prob_inner = 0.6, prob_outer = 0.80)
$folds
fold mae mae_scaled within_60 within_801 1 3.303947 0.3909302 0.8 1.0
2 2 4.442991 0.9824889 0.4 0.6
$cv
mae mae_scaled within_60 within_801 3.873469 0.6867096 0.6 0.8
Given a set of observed data including a binary response variable y and an rstanreg model of y, the classification_summary()
function returns summaries of the model’s posterior classification quality. These summaries include a confusion matrix as well as estimates of the model’s sensitivity, specificity, and overall accuracy. The cutoff
argument represents the probability cutoff to classify a new case as positive.
# Data generation
< rnorm(20)
x < 3*x
z < 1/(1+exp(z))
prob < rbinom(20, 1, prob)
y < data.frame(x = x, y = y)
example_data
# rstanreg model
< rstanarm::stan_glm(y ~ x, data = example_data,
example_model family = binomial, refresh = FALSE)
# Prediction Summary
classification_summary(model = example_model, data = example_data, cutoff = 0.5)
$confusion_matrix
0 1
y 0 8 1
1 2 9
$accuracy_rates
0.8181818
sensitivity 0.8888889
specificity 0.8500000 overall_accuracy
The classification_summary_cv()
returns the same measures but for crossvalidated estimates. The k
argument represents the number of folds to use for crossvalidation.
classification_summary_cv(model = example_model, data = example_data, k = 2, cutoff = 0.5)
$folds
fold sensitivity specificity overall_accuracy1 1 1.0000000 0.5 0.7
2 2 0.7142857 1.0 0.8
$cv
sensitivity specificity overall_accuracy1 0.8571429 0.75 0.75
Given a set of observed data including a categorical response variable y and a naiveBayes model of y, the naive_classification_summary()
function returns summaries of the model’s posterior classification quality. These summaries include a confusion matrix as well as an estimate of the model’s overall accuracy.
# Data
data(penguins_bayes, package = "bayesrules")
# naiveBayes model
< e1071::naiveBayes(species ~ bill_length_mm, data = penguins_bayes)
example_model
# Naive Classification Summary
naive_classification_summary(model = example_model, data = penguins_bayes, y = "species")
$confusion_matrix
species Adelie Chinstrap Gentoo95.39% (145) 0.00% (0) 4.61% (7)
Adelie 5.88% (4) 8.82% (6) 85.29% (58)
Chinstrap 6.45% (8) 4.84% (6) 88.71% (110)
Gentoo
$overall_accuracy
1] 0.7587209 [
Similarly naive_classification_summary_cv()
returns the cross validated confusion matrix. The k
argument represents the number of folds to use for crossvalidation.
naive_classification_summary_cv(model = example_model, data = penguins_bayes,
y = "species", k = 2)
$folds
fold Adelie Chinstrap Gentoo overall_accuracy1 1 0.9367089 0.1666667 0.8253968 0.7616279
2 2 0.9863014 0.0000000 0.9508197 0.7558140
$cv
species Adelie Chinstrap Gentoo96.05% (146) 0.00% (0) 3.95% (6)
Adelie 5.88% (4) 7.35% (5) 86.76% (59)
Chinstrap 6.45% (8) 4.84% (6) 88.71% (110) Gentoo