logistic_regression.js - epsilon

[PhilippKrone]

Hi,

I'm using natural (and thus apparatus) to perform some classification. I'm using the logistic regression classifier for various reasons. I have the issue when learning with about 500 datasets (each set consisting out of approx. 10 words), it's taking quite some time. It'd be nice to be able to set the epsilon

            if(last - current < 0.0001)
                break;
            }

dynamically when using apparatus instead of having to change it in the source code. Is there any possibility to set all those numbers via a config when calling the classifier?

Regards, Philipp Krone

[zzzwx]

+1 as well as learningRate and learning rate "division step"

[DrDub]

Why don't you write a PR for that? But if you care about performance, some changes might affect speed. You might want to test both options (with a variable vs. with a constant).