Page 188 - Data Science Algorithms in a Week

172 Luis Rabelo, Edgar Gutierrez, Sayli Bhide et al.

techniques have opened doors that increase the level of complexity in problem solving.
This has provided the environment for the renaissance of a new analytics paradigm that is
trying to deal with continuously changing environments. This new paradigm focuses on
the ability to recognize change and react quickly. For example, advanced analytics uses
continuous data sampling to provide additional insights that further enhance strategic
decisions and may assist decision makers in identifying new business opportunities
and/or new relationships, which may also support innovation and creativity (Legarra et
al., 2016). One very important aspect is the ability to forecast future perceptions and
calculate the risk of potential outcomes. The incorporation of big data capabilities can
further enhance such approaches through rich data sources and computational capabilities
that provide additional insights across a value network and/or life cycle along with real
time identification and tracking of key factors. Although big data technologies currently
exist, consensus on tools and techniques for managing and using big data to extracting
valuable insights is not well established (Gobble, 2013). Organizations are currently
trying to gain a better understanding of the new paradigm and the associated benefits
from the viewpoint of big data and advanced analytics. Complexity is always the issue.
Predictive analytics is one form of advanced analytics. Predictive analytics uses a
combination of data which may include historical, auxiliary, structured, and unstructured
data to forecast potential actions, performance, and developments. This form of advanced
analytics is considered more involved and technologically demanding than visual and
descriptive analytics. This is because predictive analytics involves statistical techniques,
AI techniques, OR/MS modeling, simulation, and/or hybrids of them to create predictive
models that quantify the likelihood of a particular outcome occurring in the future. In
addition, predictive analytics are part of systems which try to tame complexity.
Predictive analytics uses statistical techniques, AI and OR/MS modeling, simulation,
and/or hybrids. AI includes a large diverse universe of different types of techniques. The
traditional side of AI involve ontologies, semantics, expert systems, and reasoning. On
the other hand, the machine learning side of AI includes supervised, unsupervised and
reinforcement learning, including artificial neural networks, support vector machines,
deep learning, evolutionary algorithms (EAs) and other metaheuristics, and regression
trees.
Evolutionary algorithms is a family of techniques for optimization inspired by natural
evolution. Blum et al. (2012) stated that EA “is an algorithm that simulates – at some
level of abstraction – a Darwinian evolutionary system.” The most popular EAs are
Genetic Algorithms (GAs), Genetic Programming (GP), Evolutionary Strategies (ES) and
Evolutionary Programming (EP). GP is a very useful technique that has become
dominant and well developed in the last twenty years. GP is generally applicable to a
wide range of predictive analytics problems