Advanced Analytics

Business Intelligence, a mature solution for most businesses, is undergoing a transformation. BI solutions are becoming more quantitative and are taking advantage of technologies, statistical methods and data sets that are substantially different from those of the past.  These new solutions that enable technology-based decision making are generally referred to as Advanced Analytics.

Why are advanced analytics becoming more important? As it turns out, there are many questions in business, science, and other fields that can only be answered using these new techniques. In addition, the data sources available (and their size) have changed drastically – Internet-sourced Big Data requires new analytic methods to extract insights in areas such as consumer behavior. The sheer number of potential data sources (think of smartphones as portable data collection devices!) means that the quantity of data to be analyzed is enormous, and the computing power and software tools needed to analyze Big Data are fairly recent innovations.

Advanced analytic solutions address many different classes of problems.

  • One set of solutions concerns past or present events; e.g., “How did our business unit perform?”, “What sales problems are we facing right now?” etc. These solutions are sometimes referred to as Descriptive Analytics.
  • Another solution set is focused on determining the root cause of events and modeling what will happen in the future. These solutions are sometimes referred to as Predictive Analytics. This usually involves calculation over multiple variables to predict the specific outcome of some other variable. For example, a credit analysis might factor in variables such as income, payment history, loan balances, purchase history, etc. to calculate a consumer’s credit risk.
  • A third class of solution deals with testing scenarios to optimize results, such as “Which offer do consumers react to best?”, and “What portfolio mix will generate the highest return?” These solutions are sometimes referred to as Prescriptive Analytics.

Data is the critical ingredient in advanced analytics. The quality and integrity of the data will have a profound impact on the quality of the results. Advanced analytics is particularly sensitive to “noise” (i.e. missing or inaccurate data). Before starting an analytics project, there is usually significant data preparation, cleansing and transformation involved, such as coding values, grouping continuous variables, making symbolic to numeric transformations, suppressing correlated data, etc. In fact, prepping the data can take up the majority of time devoted to an advanced analytics project.

There are many different analytic techniques available today. Here’s a brief survey of some of the more important ones.

  • Data Mining helps businesses find patterns and hidden information in data sets. Typically, much of the data used is already collected in the normal course of business operations and the results are frequently used in predictive analysis scenarios. For example, an insurance company wants to predict the probability of having to pay an auto accident claim for a particular policyholder. By using data on past customers, such as type of car, frequency/cost of claims, etc., a predictive risk model can be built.
  • Affinity Analysis involves finding rules that predict the occurrence of an event based on the occurrence of other events; for example, how likely is a consumer to buy an item given than they have bought other items? This is generally used in Market Basket analysis to understand consumer buying behavior in retail. One of the key metrics in the analysis is known as Lift, which helps businesses know if statistically independent purchases are being made together more or less often than expected. This is important in understanding correlation vs. causality in the buying process. Affinity analysis, which can be computationally-intensive for large data sets, impacts decisions in marketing, promotions, merchandising, cross-selling, and up-selling.
  • Cluster Analysis helps to create structure in a data set by grouping objects, such as customers or sales transactions. The objects in a group will be more similar to each other than to objects in other groups. The cluster groups can be organized in hierarchical or partitional forms, they can be exclusive or overlapping, and the assignment of objects to groups can be complete or partial. Cluster analysis is frequently used to identify market segments, plan changes in marketing strategy, or analyze social networks. Cluster analysis is sometimes combined with Geographic Information Systems (GIS) to add a spatial dimension which can be useful in in urban planning, criminal justice, insurance, and market research.
  • Sentiment Analysis involves the sematic analysis of unstructured text to understand attitudes, opinions and emotions about products, services, marketing campaigns, political platforms, or anything else where people express opinions in an environment where data can be collected and analyzed. Sentiment analysis is frequently performed on data from social networks (where it’s sometimes known as Social Media Analysis) and blogs, but can also be performed on enterprise data such as customer support tickets. Businesses use sentiment analysis to understand brand awareness, spot product weaknesses, and find new target markets. One of the areas of current interest is mobile sentiment analysis, where location data can be collected along with sentiment, providing an additional dimension to the analysis.

The business benefits of advanced analytics are numerous. Managers can become more proactive in decision-making through the use of predictive models. By understanding trending, causation, patterns, and affinities in their markets they make better decisions and gain a competitive advantage. The applications for advanced analytics are endless, and include marketing/advertising, customer service, credit analysis, medical symptom analysis, fraud detection, crime/social issues, and many others.

GRT’s experts can help you create an analytic solution with technologies that are appropriate to and cost-effective for your business. We have experience building solutions across a wide range of security-focused industries such as banking, insurance, and energy trading. 

To learn more about how GRT can help you meet your goals, contact us.