I want to know if vectorized implementions run fast for large data set also when we are considering million features in a dataset with million training examples or at that point iteration through the training examples is a better method
Vectorized Implementation
Hey,
Vectorization is always faster than iteration, just think of it like in 1 second with iterating we can compute loss for 10 examples, and with vectorization in 1 second we can computer loss for 100 examples, because vectorization works parallely, so even if we have Million features or examples, using vectorizing is always recommended. iterations take more code and more time to execute…
Hope it is clear now 