At this point, we have a decent set of variables that can help predict whether a passenger survived the Titanic disaster. However, that data could use a bit of cleaning up in order to handle outliers and missing values. We could also try to extract other meaningful features from existing attributes to boost our predictions. In other terms, we want to do some feature engineering. Feature engineering is the key to boosting the accuracy of your predictions.
Feature engineering is the process of using domain knowledge of the data to create features that make machine learning algorithms work. - Wikipedia
ML offers what it calls data recipes to transform the data and adapt it to its linear regression and logistic regression algorithm. In Amazon ML, data recipes are part of building the predictive model, not creating the datasource...