Pattern-Based Analytics defines a powerful new approach to data analysis. It provides unique insights into the full complexity of real world data and it does this without either requiring deep mathematical skills or by requiring heroic simplifying assumptions about the important variables at work. Quantum Leap® Analyst advanced analytics products developed by Quantum Leap Innovations enable transparent, flexible discovery, visualization and analysis of informative patterns in large, complex data environments. These characteristics empower the non-statistical subject matter expert to rapidly obtain insight into their data for discovery, forecasting and decision making.
Patterns in data are prevalent across multiple domains. For example, financial market analysts often use pattern recognition to identify profitable trading opportunities. In life sciences, patterns of multi-gene associations can provide fundamental understanding of disease mechanisms as a basis for finding cures. In marketing analysis, patterns of customer behavior are fundamental to driving strategies that are customized for different customer segments. More generally, in the real world, patterns represent complex combinations of different variables or factors that drive outcomes. Patterns are a fundamental way in which we organize our experiential knowledge as a basis for decision making. The ability to discover new patterns in data can thus provide a key edge to decision makers in an ever more competitive and fast moving world.
Quantum Leap Analyst (Analyst) has some differentiating advantages over traditional statistical analytics approaches. It is based on a multi-dimensional extension of Shannon Information Theory developed by Claude Shannon, one of the founders of modern computer science. Informative patterns comprising multiple attributes can be rapidly discovered and visualized from complex, real world data. In contrast, traditional statistical methods have difficulty in identify complex, multivariate statistical associations in a transparent manner. In addition, Analyst makes no assumptions about the nature of the relationships within the data; it doesn’t require (usually unrealistic) assumptions of linear behavior but can handle arbitrarily non-linear relationships. In a similar vein, patterns can be discovered from data that can have arbitrary statistical distributions, in contrast to many traditional methods that assume normal or standard “bell curve” distributions. This latter advantage can be significant in many domains such as finance and health care. For example, in health care, biases in data gathering across a population can result in non-Gaussian distributions with “fat tails” where standard statistical analysis methods could lead to incorrect conclusions.
A fundamental characteristic of Analyst is ease of use. Patterns can be easily understood by the non-expert end user or decision maker. The traditional data-to-decision making cycle within a business environment typically involves complex data analysis performed by “quants”. The results of the analysis are then summarized in the form of reports that are more easily digested by the business end user. The cycle time associated with this process can lead to costly delays in the decision making process, as well as potentially lead to information loss during the translation of the statistics to the final report. The goal of Analyst is to remove the data-quant-end user cycle and empower the business end user to directly discover informative patterns in data as a basis for more timely decision making. Analyst can be used in a complementary fashion with spreadsheets to provide new capability to the end user.
Although pattern discovery has been employed in data analysis, it has been traditionally relegated to very specific types of analysis. For example, it has been used to discover patterns in symbolic sequences such as DNA sequences in biology. There are significant challenges to generalizing Pattern-Based discovery to diverse, heterogeneous data environments with the complications of missing data etc. that are prevalent in the real world. In addition, perhaps related to this observation, very little effort has been expended to date on utilizing the discovered patterns as a basis for more advanced analysis such as prediction and hypothesis generation. Analyst is aimed at addressing these gaps in the current state of art.
Want a demo of Quantum Leap Analyst?
Contact info@quantumleapinnovations.com for more information.
© 2013 Quantum Leap Innovations, Inc. All Rights Reserved.