KDnuggets : News : 2007 : n07 : item24 < PREVIOUS | NEXT >


Subject: MS Thesis by Tom Bergstra on Business Intelligence

Can be downloaded from

www.bergstra.com/thesis/download.php (PDF, 116 pages).

Executive summary

Business Intelligence (BI) has a notably large market size in contemporary business. Currently, most BI implementations are focussed on either generating new or replacing existing reports for decision makers. OLAP facilities are implemented in 40% of the projects and data mining is applied sporadically. The current BI market offers technology, knowledge and experience required for data mining. Nevertheless, organisations are reluctant to use data mining.

When data mining is compared with other BI methods of the BI architecture it has some intrinsic advantages: ability to make predictions about future states, ability to include estimation of likelihood of outcomes and ability to analyse a vast amount of data. Furthermore data mining - used in conjunction with other quantitative methods - enables organisations to gain a competitive advantage over their competitors. The organisational learning theory shows that data mining can retain and increase competitiveness, innovation, and effectiveness in a changing environment. Data mining should preferably be used in conjunction with other BI methods in a BI architecture.

Data mining opportunities exist for every organisation of considerable size. These opportunities include industry specific applications, or applications that can be applied across all industries. Organisations that apply data mining have often one or more of the following characteristics as well: they have a large number of customers, their personal contacts with customers are limited, they have collected a vast amount of historical data from business processes, they have current business processes that generate a massive amount of data, or have pressing reasons to use data mining.

The requirements for organisations that want to apply data mining are:

  1. high quality data in order to build accurate data mining models,
  2. software that is capable of performing all steps of data mining,
  3. hardware that is capable to store data and that provides sufficient computing power,
  4. analysts with data mining experience and skills, subject area knowledge and interest,
  5. decision makers, who are able to make decisions based on data in stead of gut feeling,
  6. demand for data mining within an organisation,
  7. senior executive(s), who support(s) data mining.

The following reasons (ordered in decreasing influence) prevent the requirements of being met: lack of executive support, no demand for data mining within organisation, lack of data mining analysts, low data quality or relevant data has not been collected and external expertise does not advocate data mining or does not have integral knowledge of data mining and business.

The adoption of data mining by an organisation is a challenging activity. It requires a paradigm shift towards a more data-based decision making style, demand for data mining has to be created and data quality has to be improved. This will take years and thus it requires a long-term view and strong commitment of senior executives. It all starts and ends with the senior executive.

Contact for feedback

MSN: tombergstra@hotmail.com
Skype: tombergstra

KDnuggets : News : 2007 : n07 : item24 < PREVIOUS | NEXT >

Copyright © 2007 KDnuggets.   Subscribe to KDnuggets News!