True or false: In order to use a neural network model, one must make sure that the ranges for the scoring data sets are within the ranges for the training data.

• A. True
• B. Not specified
• C. Sometimes
• D. False

Answer: (D)

Data preprocessing and consistent scaling are what neural networks rely on to interpret inputs. When you train, you usually compute scaling parameters (like min/max or mean and standard deviation) from the training data and apply that same transformation to every input you feed into the model during inference.

You don’t have to restrict scoring data to the exact raw ranges seen in training. What matters is applying the same preprocessing so that the inputs the model receives at inference time have the same scale as what it learned on. Raw values outside the training range can still be processed correctly after normalization, though extreme outliers can affect performance if the distribution shifts significantly or the model wasn’t trained to handle such values. So the statement is not a strict requirement; it’s false because consistent preprocessing – not matching raw ranges exactly – is what matters.