Business Intelligence solutions have been credited by analysts (see recent reports in Gartner , Fortune Magazine and DM Review ) as a forerunner in the year 2002. These solutions are believed to have been able to weather the storm of the tech-slowdown and also are predicted to do well in 2003. Given this background, for those of us interested in the management of information and knowledge it is imperative to understand the anatomy of BI solutions and what they have on offer:
Traditionally BI can be defined as a set of solutions for gathering, storing, analyzing, and providing access to data/information and to help operators make better business decisions. BI applications include the activities of decision support, query and reporting, online analytical processing, statistical analysis, forecasting, and data mining. Note this definition of BI does not distinguish between data and information, but we will come to that distinction later in this article.
In order to get a sense of the BI technology segment K-Praxis studied the following BI vendors and their offerings:
- Corporate Performance Management or Enterprise Performance Management : A range of solutions to integrate enterprise planning, strategy, budgeting, forecasting and linking them to financial reporting and financial analytics.
- Business Intelligence solutions: Solutions for gathering, storing, analyzing, and providing access to data/information and to help operators make better business decisions.
- Data Mining, Business Modeling and Business Analytics : Solutions specifically designed to mine/analyze corporate data in order to discover meaningful patterns/trends to help organizations better understand and monitor and predict organizations' strengths and weaknesses.
All of these solutions are based on two basic technologies: data mining and data warehousing. Storing and processing data through various data mining techniques enables a BI solution to crunch/mine through large databases for extraction of hidden predictive information that could provide decision-ready insights into the past and the future of an enterprise. Before analyzing the offerings of the BI vendors, let's look at some of the methodologies/processes/techniques applied in data mining:
- OLAP: OnLine Analytical Processing is a method of processing data queries in real-time in order to provide analytical insights. OLAP systems enable users to manipulate operational data using familiar business rules and terms.
- Induction and Exploratory Data Analysis: These terms refer to methods of drawing inferences based on cross-relationships that exist within the data and techniques where a number of variables are taken into account and processed for exploratory data analysis.
- Statistics: The basic statistical methods include such techniques as examining distributions of variables (for example, to identify highly asymmetric or non-normal patterns), reviewing large correlation matrices or analyzing multi-way frequency tables.
- Visualization: Through Visualization BI and Data Mining applications can present the data in a graphical or three-dimensional map, allowing the user to identify trends, patterns and relationships. Most of the BI vendors integrate visualization in their solutions.
- Neural Networks: These techniques are techniques modeled after the processes of learning in the cognitive system based on some of the neurological functions of the brain. In a away, these methods, by using brain-like processing capabilities, attempt at predicting new observations from a given set of observations.
These methods and techniques constitute the core of a BI solution and the majority of the vendors support these methods and techniques. This review is an attempt to evaluate vendors based on two criteria - what kind of view vendors take of data/information flow and its usefulness within an organization, and what type of data mining techniques/methodologies that the vendor is using to arrive at decision-ready insights.
On the count of data and information handling most of the the vendors claim that they offer a complete and comprehensive view of the enterprise data because of their ability to ferret out meaningful patterns from this otherwise un-used or under-used assets of the organization. One company which seemed to have elaborately tried to understand (but a company which is somehow fallen in bad times in the tech-slowdown) is CA and its BI solution monikered as CleverPath. CleverPath distinguishes between Data, Information, Knowledge and Action through what CA calls as Information Delivery Maturity Model . Its solution, from an information management and information flow point of view, is very much ahead of its competitors - notwithstanding its exploitation of neural network in the technology previously known as Neugents and newly-christened as CleverPath Predictive Analysis Server.
As for data mining and predictive analysis capabilities, SAS, Cognos and MircoStrategy seems to be much ahead of others - especially at their attempts to create a comprehensive Enterprise or Corporate Performance Management system based mainly on numerical data. All of these vendors offer a very elaborate enterprise predictive modeling capabilities well-equipped with a well-defined visualization tools. Informatica is now claiming that it has the most advance visualization tools based on usability and user experience.
As stated earlier most of these vendors claim that they offer what is dubbed as a 360-Degree-View of the enterprise data, but it is important to point out that most of this data is numerical data or information culled out of numbers (distinction between these entities could be another subject for analysis later on). In other words all the BI vendors deal with what is called as "structured data" data which is in the databases and excel sheets where there is huge amount of "unstructured data" which lies in emails, reports, customer transactions, marketing brochures etc which is not used by the most of the BI solutions
SAS and SPSS are unique in this respect as they have started realizing the importance of unstructured data and have created solutions on the line of other unstructured data vendors (Autonomy, Verity, Inxight, Stratify, etc). SAS's Text Miner and SPSS's recent acquisition and integration of LexiQuest is a case in point.
In conclusion, BI seems to be a must have for company dealing with huge amount of information but companies need to understand the limitations of BI as the solutions provided by them are incomplete and will require huge of amount expertise for implementation.