I want to ask why the sklearn’s fit method has no **epoch** parameter like in TensorFlow’s fit?

I think that traditional ML models like Linear Regression also need to repeatedly update their weights like DL models via an optimization process using algorithms like Gradient Descent.

So why we just specify the number of epochs for DL models but not for these ML models?

The method sklearn uses depends on how you configure the model.

If you just use linear regression, then you get an Ordinary Least Squares model.

If you want to control the learning rate and iteration, then use Stochastic Gradient Descent:

1 Like

Thank you, i get it now after some search about the “Ordinary Least Squares” for Linear regression in sklearn