1# Online
2Learners that implement the Online interface can be trained in batches. Learners of this type are great for when you either have a continuous stream of data or a dataset that is too large to fit into memory. In addition, partial training allows the model to evolve over time.
3
4## Partially Train
5To partially train an Online learner pass it a training set to its `partial()` method:
6```php
7public partial(Dataset $dataset) : void
8```
9
10```php
11$folds = $dataset->fold(3);
12
13$estimator->partial($folds[0]);
14
15$estimator->partial($folds[1]);
16
17$estimator->partial($folds[2]);
18```
19
20!!! note
21    Learner will continue to train as long as you are using the `partial()` method, however, calling `train()` on a trained or partially trained learner will reset it back to baseline first.