Do you want to access the full text of articles?

Please see our digital edition archive for the full text of articles.

Alternatively:

If you are a Chartered Accountants Ireland member, please visit the RIS service where Accountancy Ireland is available free of charge via the EBSCO databases.

If you are an Accountancy Ireland subscriber (i.e. you pay each year to receive your copy of Accountancy Ireland) please contact our Subscriptions Department quoting your subscription number and include details of the article you want.

All other users should enquire from their local public or college library about accessing full text Accountancy Ireland articles.


Data Mining - Do you really need a new IT system or just better access to your data?

Author: Rob Graham

You are either the owner of or a senior manager within a successful organisation. You know it is successful because you still exist, your auditors maintain the usual pleasantries when they visit and bank managers invite you to lunch. You either have an I.T. Department busily churning out yet more reports - or telling you why they can't - or you may have even outsourced that particular activity to a third party. So far, and maybe so good; but there is a nagging doubt somewhere in the back of your mind that it could be better and probably simpler. You cannot quite put your finger on it but somewhere along the road you feel you may have lost some of that hands on control, that made the enterprise a success in the first place. Trust that thought.

Joining the 112,000 odd visitors and exhibitors at GITEX in Dubai recently, confirmed my opinion that there has never been so much technology available to organisations as there is now. This is not necessarily a good thing.

What owners and senior managers need is not more technology but real solutions. Despite the honeyed word of some vendors, the two are not necessarily synonymous.

The next time the well heeled rep (sorry adviser) rolls up in the latest model sports version to ‘help’ you, just take a deep breath and be prepared to disappoint them. Whatever their pitch, if the information you require to direct and manage isn’t already contained within your existing system or systems then you should already be heading for the exit.

Of course the data is there, the problem is how you access it. Consider the attractions of heresy. No you do not need yet another new system but you would like an accessible, easier route to the information you require.

DATA MINING Consider the attractions of report mining and data transformation software. Any web search will bring you up to speed on currently available data mining options and will confirm the advantages of grafting such software onto your existing system(s). These include:

- Any reasonable Excel user being able to start asking new questions of existing reports within a couple of hours - Minimal to no staff re-training and zero implications for customers and other third parties. - Empowerment of managers to ask questions directly of the data - No negative impact on I.T. time - Extending the value of the investment you have made in your present system

If this business as usual approach to taking back control of your organisation appeals, then read on!

DATA CONVERSION Moving data from one system to another is never fun, and database developers know this better than anyone.

Never before have we faced such a diverse array of platforms, databases, languages, and operating systems, and this multifariousness is a key contributor to the difficulties of data conversion. At the same time, the need to quickly disseminate information globally, and across disparate computing environments, has never been greater.

By learning a few data conversion tricks and techniques, you can more readily manage your next data conversion project. Data conversion projects consist of three related processes (ETL):

- extraction, - transformation, and - loading .

During the first process, programmers write custom code or use specialized tools to extract data from a source database. Next, the collected data is transformed into formats suitable for the new application, perhaps a data warehouse. Finally, the transformed data is loaded into the target database. Report mining can help you perform these tasks in less time, and with fewer difficulties, than traditional methods can.

For several reasons, extracting data is a time-consuming, error-prone process. First, programmers must understand source database schemata before requisite data can be extracted. In the case of relational databases such as Oracle, DB2, SQL Server, and Informix Foundation, this can be a daunting task because such systems are for normalised schemata that encode data and store it in multiple tables.

To extract normalised data, programmers must understand precisely how the data is encoded and how the tables are related. Even with excellent documentation, learning the vagaries of normalised schemata is challenging.

Of course, not all databases are normalised. Older legacy platforms, most notably mini-computers from the 1970s and early 1980s, often store data in proprietary data structures, requiring programmers to have platform-specific knowledge of these structures. In either case, programmers must also understand the host platform's operating system, and be conversant with a supported programming language or ETL tool.

Finally, programmers must understand the context of source data. For example, let's suppose a field in a source database contains sales values. Programmers must know whether the figures represent gross or net values, and whether the values are in a foreign currency or not. One misunderstanding can undermine an entire conversion effort. Obviously, extracting data is a difficult, time-consuming process.

Report mining tools can help mitigate both the complexity and time needed to perform data extraction because they let programmers use spooled files instead of databases as the data source. Spooled files are electronic versions of reports before they are printed on paper. To generate spooled files, report programs extract data from databases, and then present that data in an intuitive format. In this regard, spooled files mask the complexities of underlying database schemata.

Using spooled files lets programmers access and extract data without needing to understand a source platform's operating system, application, and database. And because virtually every application on every platform generates spooled files, report mining tools function as universal passkeys to data. Best of all, report mining tools let programmers extract data directly from spooled files without programming.

The future control of your enterprise can be just a web search away. After all, don't you remember the many joys of actually being in control? ‘Pass’ on yet more complicated technology and benefit from a real solution.

Rob Graham is Product Manager with Datawatch-Europe Ltd.