Jump to content
Powered by

Digging for data: looking for the essential

The data-mining tool KNIME has no difficulty in discovering key relationships within large amounts of data. The software stands out thanks to three particular features: it is based on the open source development method, it integrates current research methods and is doted with an intuitive operational concept. KNIME was presented to the audience at the recent Innovation Forum in Heidelberg.

Prof. Dr. Michael Berthold (Photo: do it.online)
A pharmaceutical company wants to analyse the data from a drug test. A publishing house wants to choose from millions of addresses those clients who respond to publicity material. An ENT doctor wants to test thousands of surgical data relating to the effectiveness of medical interventions. All this is possible - the KNIME (Konstanzer Information Miner) platform gives users the possibility of analysing and presenting their data according to their specific needs.
“KNIME, pronounced [naIm], is a modular platform for the interactive exploration of data,” explains Prof. Dr. Michael Berthold from the Department of Bioinformatics and Information Mining at the University of Constance. In short, KNIME answers questions on the basis of existing data. “The goal is to find interesting relationships within large amounts of data.” The keyword is data mining.

Interactive foraging and statistical data analysis

The individual analysis steps are visible in workflows. (Photo: do it.online)
The individual analysis steps are visible in workflows. (Photo: do it.online)
Working with colleagues in his university department, Michael Berthold developed this platform that has been in use since 2004. In 2005, Berthold founded KNIME GmbH. The close connection of the company with the university and close cooperation with research-intensive companies has, right from the very start, ensured that new industrial and research findings and experience have been included in the development of KNIME. The platform is based on open source, which means that clients can easily integrate their own programmes.
This is particularly important for life science clients who habitually work with a large number of specialised programmes. In addition, they can save on expensive licensing costs.

KNIME’s broad range of functions enables an understanding of data on many levels – from the simple interactive foraging of data up to professional statistical data analysis. In addition, KNIME is easy to understand and can be operated intuitively. One user is able to visually create data flows from the individual analysis steps, making it possible for another person to be able to understand the process at a glance. New workflows can be created with simple drag & drop. The data will be automatically transferred from one analysis step to another and the intermediary results can be investigated at any time and used for additional operations.

Used for large amounts of data

A large number of KNIME’s clients work in the life sciences field, but clients also include credit institutes and publishing houses. In principle, KNIME can be used for all kinds of large data sets, for example in client segmentation and process optimisation. An interactive decision tree supports users in decision-making processes.

A current example from the banking sector ably demonstrates the advantages of the software. The original issue raised by this particular bank related to the possibility of broadening its client base; but data analysis with KNIME showed that it was better to focus on a few exclusive clients. “A modular platform like KNIME can focus its analysis very quickly on an issue that is different from the original question,” said Michael Berthold. The prerequisite for significant analyses is the preliminary work: the structuring and tidying up of data. This is very work intensive but once done, other questions can also be answered easily and quickly.
An interactive decision tree supports users in decision-making processes. (Photo: do it.online)
An interactive decision tree supports users in decision-making processes. (Photo: do it.online)
Michael Berthold has many plans for the KNIME platform, for example expanding the functions of KNIME and uncovering new fields of application, including business intelligence (BI) and customer relationship management (CRM). In addition, he intends to introduce a certification service in order to guarantee uniform quality for the commercial expansion of his numerous industrial software partners.

Source: doit.online - 19th March 2008
Website address: https://www.gesundheitsindustrie-bw.de/en/article/news/digging-for-data-looking-for-the-essential