A SAS data scientist is preparing features for a neural network in PROC HPNEURAL. Input variables have widely different scales: 'age' ranges 18-90, 'income' ranges 20000-500000, and 'loan_amount' ranges 1000-50000. Why is standardization critical before training?
Select an answer to reveal the explanation.
Short Explanation and Infographic
Here's the deal — b is correct because neural networks use gradient-based optimization. When input variables have vastly different scales, the weight gradients have very different magnitudes, causing the optimizer to take large steps along some dimensions and tiny steps along others.
Full explanation below image
Full Explanation
B is correct because neural networks use gradient-based optimization. When input variables have vastly different scales, the weight gradients have very different magnitudes, causing the optimizer to take large steps along some dimensions and tiny steps along others. This leads to slow convergence (oscillation in the loss landscape) or numerical instability. Standardizing inputs to similar scales normalizes the gradient landscape and enables stable, efficient training. A is wrong because PROC HPNEURAL accepts continuous values in any range. C is wrong because standardization produces continuous values, not nominal categories. D is wrong because the output activation function (not input scaling) determines the output range.