There is a grey area between cleaning data and feature engineering. At some point, the more sophisticated methods for data cleaning becomes imputation and infer

There is a grey area between cleaning data and feature engineering. At some point, the more sophisticated methods for data cleaning becomes imputation and inference of features. I wonder what are the ramifications of this.