Question

Discuss the concept of data-driven and model-driven DSS. Illustrate your answer both by discussion of long established examples of these classes of system and by discussion of the role of newer developments such as Business Intelligence.

Answer 1

What are Data-driven DSS?

Typical questions answered by a data-driven DSS:

• How many washing-up liquids were sold in total in December 2003?
• Compare the turnover of our washing-up liquid in St. Gallen to the turnover of this product in Geneva for the last four months.
• Which washing-up liquid is favoured by business women over fifty who buy the magazine Gala regularly?

This also explains the two sub-types of data-driven DSS:Such questions can be answered by a data-driven DSS. Data-driven DSSs help prepare decisions by showing developments in the past and by identifying relations or patterns.

• The first sub-type OLAP provides the means for interactive ad-hoc exploration of huge amounts of data to gather the information necessary for the decision.
• The second one, Data-mining, provides the means for automatically find structure in the form of relations and/or patterns in large databases. These can then be used as models of the data to understand the data’s structure (e.g., what item is usually bought together with spaghettis), make predictions (what is the probability that a customer who buys spaghetti also buys red wine), find a most likely explanation (given the shopping of this person she is likely to have a dinner party tonight).

The aim of data-driven DSS is to give the user a simple access to large amounts of data from various sources (both internal and external) on which he or she can run real-time queries. A data-driven DSS helps to transform large data sets into meaningful, easy to read business information.

Even if a data-driven DSS has models, they are in general very simple and only used to compute information like averages or totals.

What are Model-driven Decision-Support Systems?

Typical questions answered by a model-driven DSS:

• What will happen when we shift the production abroad?
• Which impact will the lower wage costs have on attracting our investors?
• Which impact will the increase of the working week to 42 hours have on our productivity?

By interacting with a model-driven DSS, the user primarily accesses and manipulates a model like statistical, financial, optimization , or simulation models. The model allows the user to perform primarily mathematical analysis on both internal and external data. But in general, a model-driven DSS doesn’t need large amounts of data.Such questions can be answered by a model-driven DSS. It helps to understand the impact of decisions on organizations.

A model-driven DSS is primarily used to analyze a situation with a well-defined and structured problem, where assumptions on causalities can be assessed. Mostly a model-driven DSS is used for what-if-analysis.