1# Probabilistic
2Estimators that implement the Probabilistic interface have the `proba()` method that returns an array of joint probability estimates for every possible class or cluster number. Probabilities are useful for ascertaining the degree to which the estimator is *certain* about a particular outcome. A value of 1 indicates that the estimator is 100% certain about a particular class or cluster number. Conversely, a value of 0 means that the estimator is 100% certain that it's *not* that class or cluster number. When the probabilities are considered together they are called a *joint* distribution and always sum to 1.
3
4## Predict Probabilities
5Return the joint probability estimates from a dataset:
6```php
7public proba(Dataset $dataset) : array
8```
9
10```php
11$probabilities = $estimator->proba($dataset);
12
13var_dump($probabilities);
14```
15
16```sh
17array(2) {
18	[0] => array(2) {
19		['monster'] => 0.975,
20		['not monster'] => 0.025,
21	}
22	[1] => array(2) {
23		['monster'] => 0.2,
24		['not monster'] => 0.8,
25	}
26	[2] => array(2) {
27		['monster'] => 0.6,
28		['not monster'] => 0.4,
29	}
30}
31```
32