Ask the Experts: Approaches to Data Mining ERP

  • Written By:
  • Published:

From one of our readers comes this question:

I am a student of IT Management; I have an ERP course and I am supposed to write an article to review new aspects of ERP systems. I’ve decided to explore the reasons for using data mining techniques in ERP systems—and to look at different modules to which these techniques have been applied. I am going to prepare a framework to determine which ERP vendors use data mining techniques and whether these techniques are more effective in particular modules.

I would be thankful if you could provide me with any kind of information related to this research. I look forward to hearing from you soon.

We submitted this to our analyst team—here’s what they had to say:

Overview: Data mining is not just the act of choosing the right reports by whatever breakdown an ERP vendor has provided. Data mining means using data that has been collected for analytics—that is, looking at patterns, trends, and opportunities, and even going further to provide feedback in the business processes.

Data mining in ERP (which can also be considered business intelligence [BI]) is a very large topic and is very module- or industry-specific. In banking, for example, data mining is used in real time to determine patterns of credit and debit card transactions, looking for fraudulent use. It is also used to search for anomalies in funds transfers, again mainly looking for fraud. Data mining in banking is used to look for out-of-the-normal business transactions such as bad debt protection, to examine historical data to search for market trends, to perform reviews of service profitabilities, and more.

In sales, it is used to review the performance of salesmen, customers, buyers, and vendors, as well as product sales, warrantees, markets, and product profitability.

Every industry has some need for data mining. At the technical level, the building of data cubes lends itself to presenting information that is going to be statistically analyzed, based on collections of information from the ERP system. At the business level, which drives the technical level, information is analyzed to predict profitability, customer service, or safety.

In a retail environment, one can look at the sales of a product by date range, store, city, province, etc. It’s essentially a report. The data can be saved in a cube so that one does not have to rerun the report. Now, one would like to go further and examine the data associatively, or analytically. As examples, for anyone who bought Product X, what else did they buy? And (clustering) what else did similar shoppers buy? Call that the basket. Then the business can make recommendations (as does Amazon to its customers) based on the fact that other customers also bought products Y and Z.

In a more generalized view, data mining in ERP can be divided into several perspectives, including industry types, and compliance and regulatory issues. There are elements of data mining in ERP that encroach upon different knowledge bases.

Usage: Within the context of the end use of the captured data, there are broad implications in regards to ownership of data (i.e., public domain versus private domain) and subsequent legal and ethical considerations as to whether the data collected can be sold and redistributed to a third party.

Strategic business uses of data mining: As an end result of organizations implementing ERP systems and CRM, vast amounts of data were accumulated. The information collected required storage (data warehouses, data marts) and a method to retrieve, extract, and manipulate the data into a manageable form for the purposes of analytics. The end result was the creation of a field known as business intelligence. Through the use of BI, both public and private sector organizations could apply techniques (data mining) to identify trends, demographics, consumer preferences, and patterns, and to orient advertising of a product or service to the relevant population segment.

Some known uses of data mining are:

  • fraud detection

  • quality defect analysis

  • supply chain management (SCM)

  • focused hiring

These are just a few examples of data mining, yet there are other ways in which data mining applications can prove useful to organizations. The value that data mining and business intelligence represent to an organization is the ability to identify trends and shifts in demand patterns. This can help organizations reduce costs and benefit from increased sales revenue opportunities.

Here’s my take on data mining relating to health care compliance for electronic data. On April 14, 2006 the Health Insurance Portability and Accountability Act (HIPAA) took effect. HIPAA is a set of guidelines that US health care organizations must follow to the letter when dealing with electronic media such as electronic medical records, medical billing, and patient accounts. HIPAA ensures that these organizations protect the integrity of their patient data—which is the life line of the health care industry.

There are three levels of security that must be enforced at all times. They are

  • administrative security - assignment of security responsibility to an individual

  • physical security - required to protect electronic systems, equipment, and data

  • technical security – relates to the authentication and encryption used to control access to data

(See for more information about HIPAA)

Since patients put their sole trust in the health care industry to keep their information confidential, the onus is on health care organizations to ensure that patient data is indeed secure; if a health care facility doesn’t follow HIPAA guidelines, it is compromising patient data and there could be substantial legal ramifications.

Data mining to review illnesses and effectiveness of treatments is providing discoveries into new medical treatments, and helping to eliminate ineffective ones.

In regards to data mining for ERP, many firms are turning to what is known as manufacturing analytics. What this represents is a BI solution layer on top of traditional manufacturing technologies, enabling users to extract data from their manufacturing environment.

What does this mean exactly? Let’s say, for example, that a manufacturer produces cars and that it must procure car parts from multiple suppliers. If these components do not arrive on time, this will negatively affect their production runs, essentially decreasing the company’s bottom line.

By mining data in the ERP system, the manufacturer can

  • distinguish which suppliers negate on delivering their components on time (and thus, in turn, decide to go with another supplier)

  • decide if different manufacturing processes can be performed before the components arrive, so as not to waste time in the manufacturing environment

Data mining in an ERP environment can also help to lower costs within the financial section in the ERP, as well as help HR managers extract important information about employees in the organization. For example, if a particular skill is needed for a specific manufacturing task, the HR manager can quickly pull up this information and make an informed decision on where the employee’s skills would be best suited for that task.

Over the past decade, the Internet has changed the way organizations do business. With the advent of the Internet, transactional data has reached enormous proportions. Let’s face it, data is perpetual. So the question is: how can businesses use their transactional data to their advantage?

Data mining (the process of retrieving information) is vital to understanding what’s going on within the business—how it’s doing financially, where problems lie, and where improvements can be made. As such, it’s important to be able to capture data in a secure and effective manner in order to make strategic business decisions.

The volumes of data that businesses deal with daily, the strict compliance regulations (e.g., SOX) they must adhere to, as well as the organizational operational policies imposed on them, are just a few of the things fuelling the need for data controls.

The newest trend in data mining is continuous controls monitoring (CCM). CCM’s powerful analytic capabilities enable organizations to gain immediate insight into the transactional data underlying their business and financial reporting processes.

Here is an example of where CCM is used:

A CCM interface to the inventory holdings in a distribution business is used to review potential stock-outs for thousands of stocked items. Using one specific case, the system recognized that according to projections, in 10 weeks that product would be out of stock, and since the lead time for replenishment is eight weeks, the CCM system output determined the optimal ordering method, the optimal quantity, and prepared a purchase order for a buyer to approve.

What makes CCM different? By continuously and independently analyzing financial transaction data, CCM applications can verify and validate the data against the organization's control parameters and business rules. CCM benefits busy organizations by rapidly analyzing large volumes of data, identifying suspicious activities, and providing alerts should any breaches in security occur. This functionality enables users to detect the specific transaction responsible—before the problem worsens. It then automatically stores the data being analyzed—resulting in information that can be easily browsed and instantaneously retrieved.

TEC-certified software and services for data mining/business intelligence (BI):

Bitam - Bitam Artus
InetSoft Technology - InetSoft Suite
LogiXML - Logi 9 Business Intelligence Platform
Oracle - Hyperion System 9
Compare BI solutions side-by-side

Do you have a question you'd like to submit to our panel of experts? It's easy--send us your questions, we'll respond and publish selected questions.
comments powered by Disqus