Momocs (version 1.3.2)

classification_metrics: Calculate classification metrics on a confusion matrix

Description

In some cases, the class correctness or the proportion of correctly classified individuals is not enough, so here are more detailed metrics when working on classification.

Usage

classification_metrics(x)

Arguments

x

a table or an LDA object

Value

a list with the following components is returned:

1. accuracy the fraction of instances that are correctly classified

2. macro_prf data.frame containing precision (the fraction of correct predictions for a certain class); recall, the fraction of instances of a class that were correctly predicted; f1 the harmonic mean (or a weighted average) of precision and recall.

3. macro_avg, just the average of the three macro_prf indices

4. ova a list of one-vs-all confusion matrices for each class

5. ova_sum a single of all ova matrices

6. kappa measure of agreement between the predictions and the actual labels

See Also

The pages below are of great interest to understand these metrics. The code used is partley derived from the Revolution Analytics blog post (with their authorization). Thanks to them!

1. https://en.wikipedia.org/wiki/Precision_and_recall

2. https://blog.revolutionanalytics.com/2016/03/com_class_eval_metrics_r.html

3. https://www.r-bloggers.com/2016/03/is-your-classification-model-making-lucky-guesses/

Other multivariate: CLUST(), KMEANS(), KMEDOIDS(), LDA(), MANOVA_PW(), MANOVA(), MDS(), MSHAPES(), NMDS(), PCA()

Examples

# NOT RUN {
# some morphometrics on 'hearts'
hearts %>% fgProcrustes(tol=1) %>%
coo_slide(ldk=1) %>% efourier(norm=FALSE) %>% PCA() %>%
# now the LDA and its summary
LDA(~aut) %>% classification_metrics()
# }