Clarification on how fit/predict work together?

[winknotes]

I think I have a conceptual problem understanding these two commands but in my mind you'd build a model with fit and somehow use that model to forecast (predict) future events right?  But for the life of me I can't find any examples of this in practice.  To see it in pseudo-SPL might be something like this:

| mstats avg(some_metric) avg(some_other_metric) avg(yet_another_metric) WHERE index = my_index span=1d
| table some_metric some_other_metric yet_another_metric _time
| fit Ridge some_metric from some_other_metric yet_another_metric into my_model
| predict some_metric as prediction 

I guess everything I read and examples I see treat fit and predict as exclusive commands.  So I guess I have a couple questions.  

What's the point of a model created with fit?  
Do fit and predict work together to forecast?  

Thank you to anyone who can clear up my thinking on this.   

[skoelpin · Accepted solution]

The `fit` command is used to create a new model. This can be used for regression, classification, clustering etc.. Once a model is created with `fit`, you can then use that model with the `apply` command. The `predict` command does not require either the `fit` or the `apply` command as it works out of box. The `predict` command is used more for linear regression problems where you would find the slope and extrapolate out into the future.