What type of target can the MBR Memory Based Reasoning Node predict?

The Memory-Based Reasoning (MBR) Node in SAS Enterprise Miner is designed to handle both categorical and continuous targets. This flexibility is a significant feature of the MBR method, allowing it to be applied in a variety of predictive modeling scenarios.

When dealing with categorical targets, the MBR approach functions by finding the closest historical records (neighbors) in the training dataset and using them to make predictions based on the modal value (commonly used class) of these neighbors. This is particularly useful for classification tasks where you need to predict discrete categories.

In the case of continuous targets, the MBR node predictively averages the values associated with the nearest neighbors in the dataset. This process is effective for regression tasks where the outcome is a continuous variable.

The ability to predict both types of targets makes the MBR node versatile and applicable across different modeling tasks, enhancing its usability in diverse analytical environments.