Learning From Data: Building, Evaluating And Understanding Models
Free (open access)
Andreas S. Weigend
of the Invited Technical Conference Analyzing tick-by-tick market data in real time, uncovering trading styles and understanding their profit and risk characteristics, managing portfolios of thousands of securities, and flagging potential fraud in millions of daily transactions are some of the challenges now sweeping the financial industries. Knowledge discovery and data mining approaches address these challenges and try to extract previously unknown, valid and comprehensible knowledge from large data sets. These approaches have emerged from several historically disjoint communities that include artificial intelligence, neural networks, evolutionary programming, reinforcement learning, signal processing, decision science, statistics, econometrics, probabilistic modeling and computational learning.