March 2021
Beginner to intermediate
498 pages
12h 3m
English
Ontology recapitulates philology.
–Willard Van Orman Quine (c.f. Ernst Haeckel)
In a classification model, there are numerous metrics that might express the “goodness” of a model. Accuracy is often the default metric used, and is simply the number of right answers divided by the number of data points. For example, consider this hypothetical confusion matrix:
|
Predicted/Actual |
Human |
Octopus |
Penguin |
|
Human |
5 |
0 |
2 |
|
Octopus |
3 |
3 |
3 |
|
Penguin |
0 |
1 |
11 |
There are 28 observations of organisms, and 19 were classified accurately, hence the accuracy is approximately 68%. Other commonly used metrics are precision, recall, and F1 score.
Related concepts: F1 score, precision, recall
Read now
Unlock full access