The SAFE Method for Feature Reduction and Prediction

In the age of data availability, being able to analyze high-dimensional data is crucial for decision-making. Nevertheless, accounting for a large number of features can introduce several challenges and dimensionality reduction techniques are necessary to improve both the speed of the machine learning algorithm as well as its accuracy. To address this concern, in this research we develop a new dimensionality reduction algorithm called Supervised Approach for Feature Engineering (SAFE). Our proposed approach identifies a projection that best explains, not the variance in features, but how well those features are correlated with the dependent variable, facilitating any future supervised machine learning algorithms.