State-of-the-Art in Performance Metrics and Future Directions for Data Science Algorithms

Abstract

With the advancement in cutting-edge technology, a huge volume of data is generated every day. Data science is one of the unique and most motivating area of research which is progressively popular now a days. Data science plays a vital role for analysing a massive volume of data that cannot be processed by conventional technologies in real time. Its aim is to find solutions from existing data in order to improve our existing systems to reduce time and make it cost effective. In this article we explored why we need data science, big data and data mining techniques. It gives readers clear intuition towards the basic steps required for dealing with data science and analytics problem. This work focuses more on the supervised learning. In this review we review all related articles in the field of Healthcare that needs an improvement for making our healthcare systems more reliable. This also highlights an introduction of important techniques of supervised learning in the domain of healthcare for better understanding of techniques. In particular Recommendation regarding the choice of suitable activation function and evaluation metric to improve the performance of classifiers is explained briefly.