C1_W3_cost-function-for-logistic-regression_log-base-what

Course Q&A Machine Learning Specialization Supervised ML: Regression and Classification
1

In the video of Week 3 (cost-function-for-logistic-regression), the log function is being used.
Is it base e, 2 or 10?

2

Hello @adnan_h_mohamed, it is base-e, and that’s why when you differentiate it to get the gradient, there is no additional constant.

Popular implementations (sklearn, tensorflow) also use base-e.

Having said that, there is no harm to use base-2 or base-10 as either of them will just introduce a positive constant to the gradients, but we usually use base-e.

3

Thanks for the clarification :blush:

4

Your are welcome @adnan_h_mohamed!