What is NOT a feature of decision predictions?

The focus on decision predictions typically involves the classification and categorization of data, especially with respect to identifying patterns or making decisions based on categorical target variables. Features such as fraud detection and item classification fall within this domain, leveraging algorithms that utilize historical data to predict outcomes for new instances.

In contrast, decision predictions are not inherently related to time-series data. Time-series analysis is concerned with data points collected or recorded at specific time intervals, aiming to forecast future values based on past trends. While decision predictions can sometimes be applied within a time-series context, they fundamentally differ in that they primarily deal with analyzing categorical outcomes rather than temporal changes or trends over time. Hence, the statement about time-series data does not align with the core features associated with decision predictions, making it the correct choice for what is not a defining characteristic in this context.