Under which situations will you prefer to go for "Stochastic gradient descent" and give recommendations accordingly?Select an option Clear ResponseCalculations are done in such a way that the training data instance and its updates are calculated immediately.Calculations are done in such a way that the training data instance and its updates are calculated as a batch process.Calculations are done in such a way that the training data instance and its updates are calculated every day at a particular time.Calculations are done in such a way that the training data instance is calculated immediately, but its updates will happen in a batch.

The Stochastic Gradient Descent (SGD) method is preferred in situations where calculations are done in such a way that the training data instance and its updates are calculated immediately. This is because SGD updates the parameters for each training example one by one, as opposed to batch gradient descent which calculates the error for each example in the training dataset, but updates the model only after all training examples have been evaluated.

One of the main advantages of SGD is that it's computationally faster, especially when dealing with large datasets. This is because it only needs to use one training example at a time. This also allows SGD to be used for online learning models, where the model is updated on-the-fly as new training examples come in.

However, SGD also has some disadvantages. It's more susceptible to noise in the training data and can lead to a less stable convergence. It may also require more hyperparameter tuning, such as the learning rate and the number of iterations.

So, in summary, you would prefer to use SGD when