Feature Engineering Tools and Techniques for Better Classification Performance
Feature engineering has been the focus of interest for some time and it is still limited or under studied. Therefore, more determined attempts are required to help forward feature engineering process in the context of learning algorithms to predict better results and behaviours. With the huge amount of data available and the consequent requirements for Artificial Intelligence and good machine learning techniques, new problems arise and novel approaches to feature engineering techniques are in demand. This paper presents a comprehensive survey of methodologies, tools and techniques used for feature engineering with the purpose of improving model(classifier) accuracy on unseen data and also presents applications of feature engineering in text classification, clinical text classification, link prediction on social networks, knowledge base construction, fraud detection and other domains, used to achieve high performance of predictive learning algorithms in terms of model accuracy.