How many documents must be coded with each choice for an active learning model to build?

The correct answer is that an active learning model requires at least five documents to be coded for effective functioning. Active learning is a machine learning technique where a model is trained incrementally and selectively queries the user or expert to label data points that are most informative for the learning process.

Starting with a minimum of five coded documents allows the model to establish a more reliable baseline for understanding the patterns and relationships in the data. This foundational set of coded documents helps the model learn the distinguishing features and nuances within the data. If less than five documents are coded, the model may not achieve the necessary diversity in the labeled inputs, potentially leading to overfitting or undertraining, which could undermine its ability to generalize effectively to unseen data.

In project management and data science practices, having more than five coded documents can enhance the model's ability to accurately identify and prioritize what additional data should be labeled next. This creates a more robust cycle of learning and improves the overall effectiveness of the active learning framework.

The other options, being lower numbers, do not provide sufficient data volume to establish a reliable model foundation, which is critical for the success of active learning strategies.