If the dependent variable is numeric, the neural network is typically used for

• A. Classification
• B. Dimensionality reduction
• C. Regression
• D. Clustering

Answer: (C)

When the target variable is numeric, you’re predicting a continuous value from inputs. A neural network this setup learns a mapping to real numbers and is trained with regression losses such as mean squared error or mean absolute error. The output layer is typically linear, allowing any real-valued prediction. This is different from classification, which predicts categories and uses probability-like outputs (sigmoid or softmax). Dimensionality reduction and clustering are not supervised prediction of a numeric target; they serve other purposes like reducing features or grouping similar data. So the task best described here is regression.