BigDataFr recommends: Understanding Dimensionality Reduction and its Applications
Dimensionality reduction as means of feature extraction
Feature extraction is a very broad and essential area of data science. It’s goal is to take out salient and informative features from input data, so that they can be used further in predictive algorithms. Modern data scientists observe large amounts of data, which is hard to process at once: data can be raw, unstructured, high dimensional, or noisy. Thus, extracting salient features is vital for successful applications of machine learning algorithms. Feature extraction is a widely discussed research topic.