BACKGROUND: BG Technology is the research and development business unit of the BG Plc. Group (formally British Gas). BG operates around the world across the whole of the gas chain, from exploration and production through transmission and distribution to the gas markets. BG Technology is also involved in a number of collaborative research projects, funded by governments or organizations such as the European Union. BG Technology has been using Clementine since 1995 in technical applications and as a tool to support new research.

PLATFORMS: Clementine runs on UNIX, Windows NT or on a Windows 95 client connected to a server. We run the system at BG Technology on Windows NT 4.0.

PROBLEM SOLVED: Clementine has been used in several projects to support research programs for various parts of the business. One of the projects in 1998 involved the analysis of gas demand forecasting data. Clementine was used to optimally split demand forecast data into two classes (summer and winter) and to represent different relationships between weather and demand. It was also used to investigate use of wind direction as an input variable and to identify relationships between model forecast errors and weather conditions.

PRODUCT FUNCTIONALITY: Clementine provides seven modeling tools including neural networks, clustering and rule induction algorithms. There are also pre-processing and cleansing facilities and interactive visualization tools, all accessed through an easy-to-use visual programming front end. The Clementine External Module Interface provides easy access to other tools, algorithms and applications. Clementine reads data from any ODBC-compliant database, flat files, MS Excel, SAS, Oracle Express and SPSS.

STRENGTHS: Clementine is versatile and easy to use. The graphical and modular front end makes it accessible to non-IT specialists, while providing facilities for advanced users to customize the components and environment as required. Hypotheses and new applications can be explored quickly while still providing the power to perform in-depth detailed analysis.

WEAKNESSES: Clementine draws data from relational databases and other sources, then carries out most of its processing in memory. This means Clementine's performance is limited by the specification of the machine on which it runs. To address this problem, ISL is developing a middleware tier (CHESS ­ the Clementine High-Efficiency Scalable Server) which, as far as possible, carries out data mining operations within the host DBMS. CHESS is scheduled for release in the first quarter of 1999. The deployment facilities are also limited ­ trained models can only be exported in C-code. This will be remedied in the Clementine Solutions Exporter, which can export a full set of operations for other applications.

SELECTION CRITERIA: Clementine has a proven record within BG Technology as a versatile and efficient data mining tool. It has open data access and is easily integrated with other components. We chose Clementine to help solve our problem because it allowed us to explore several options quickly while still performing in-depth analysis. The graphical-based interface made it easy for forecasting experts as well experienced data miners to visualize the findings.

DELIVERABLES: We continue to use Clementine for a number of business projects, improving the use of our data and the effectiveness of our research programs. Season-specific models were produced as a result of this project. We are also exploring some of the emerging aspects of data mining and have recently finished a project that demonstrates the potential of text mining which integrate information extraction with data mining.

VENDOR SUPPORT: We have enjoyed a rewarding partnership with ISL. They provided us with several excellent training courses in Clementine and data mining. We are also involved in the Clementine User Group (CLUG), which produces regular newsletters, gives master classes and holds an annual conference. The on-line and telephone support has always been fast and efficient.

DOCUMENTATION: ISL provides a comprehensive manual and user guide for Clementine. Other resources include case studies, FAQs, training material and white papers on how to mine data for various applications.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access