What does the Metadata Advisor automatically reject when using the default advanced options?

The Metadata Advisor in SAS Enterprise Miner is designed to assess variables in a dataset based on their quality and usability for modeling purposes. When using the default advanced options, the advisor automatically rejects variables that have more than 50% missing values. This approach is grounded in the idea that if a variable is missing too much data, it may not be reliable or useful for analysis. High rates of missingness suggest that the variable could introduce bias or uncertainties into model predictions, ultimately impairing the model's performance or interpretability.

By focusing on this threshold of 50% missing data, the Metadata Advisor ensures that the variables retained for analysis are more likely to contribute meaningfully to the predictive modeling process. Variables with fewer missing values generally provide more robust information and are preferable for creating reliable models. This automated rejection process helps streamline the data preparation phase, allowing analysts and data scientists to focus their efforts on the most pertinent and trustworthy variables.