The classification you have seen and experienced so far is a two-class classification where the target variable can be of two classes. In multiclass classification, you classify in more than two classes, for example continuing on our hypothetical tumor problem, for a given tumor size and age of a patient, you might predict one of these three classes as the possibility of a patient being affected with cancer: High, Medium, and Low. In theory, a target variable can have any number of classes.
ML Studio lets you evaluate your model with an accuracy that is calculated as a ratio of the number of correct predictions versus the incorrect ones. Consider the following table:
|
Age |
Tumor size |
Actual class |
Predicted class |
|---|---|---|---|
|
32 |
135 |
Low |
Medium |
|
47 |
121 |
Medium |
Medium |
|
28 |
156 |
Medium |
High |
|
45 |
162 |
High |
High |
|
77 |
107 |
Medium |
Medium |
The following can be the evaluation metrics where in the columns, the text is marked in bold...