An ASP Approach to Data Mining

If digiMine gets its way, data warehousing and data mining will become legitimate outsourced services along the same lines as e-mail and conventional application hosting.

We believe digiMine’s ASP-based approach to data warehousing and data mining makes a heck of a lot of sense and can provide compelling value for most organizations – the variety and the granularity of the reports that digiMine’s data-mining operation can produce will certainly be a boon to the marketing and sales efforts of any organization. We can think of several potential drawbacks, however, that might also be associated with it.

First, digiMine hosts its data-warehousing and data-mining services almost exclusively on SQL Server 2000. Although this has the advantage of simplifying administration from digiMine’s point of view, it also restricts the options available to customers who could benefit from solutions that don’t necessarily run on Windows 2000 – or to customers who don’t maintain digiMine’s requisite OLEDB-compliant data repositories in-house. (They are out there, by the way, and they usually take the form of niche or proprietary and legacy solutions.)

Second, with a $5,000 first-time setup fee and with monthly fees from $7,500 on up, digiMine’s price tag – through a fraction of what it might cost to roll out and maintain your own in-house data-warehousing operating – is not inconsiderable.

Background Information

DigiMine’s twist on the traditional outsourcing theme is a compelling one: Not only will it host your mission-critical business data, but it will also sift through your data with its data-mining and online-analytical-processing (OLAP) tools and compile a variety of reports for you. Viola – instant business intelligence.

In this sense, digiMine proposes to effectively build, design and host a data warehouse for your business. Company officials acknowledge that outside of specific vertical markets, data warehousing hasn’t exactly caught on like wildfire. They say the reason for this is that data warehouses are inordinately difficult to tune and configure in the first place.

Because digiMine addresses potential configuration issues on its end, and because it can automatically generate any of several different prepackaged report types, it purports to do away with this issue altogether. Company officials claim that customers should be up and running and receiving valuable business intelligence within a few weeks – as opposed to the six-month lead time that it might take a company to develop, configure and tune a data warehouse on its own.

Pulling It Through Its Paces

For the purpose of our test, we used an agglomeration of sample data that digiMine provided. We wanted to test digiMine’s ability to mine data for information and coherently render it within the context of a report. DigiMine’s canned data let us do that.

You can send digiMine the data that you would like warehoused, or you can upload it (encrypted) over an Internet connection. Once the data is uploaded, digiMine uses a mechanism (called a “data slurper”) to periodically refresh your data. Data slurpers reside on you internal data repositories (digiMine claims that they’ll work with any OLE-DB-compliant data store), grabbing relevant information – such as user profiles, Web logs and catalof and SQL Server transactional data-encrypting it – and sending it over the wire. Caveat emptor. Because we were using a sample pool of data provided by digiMine, we were unable to test the performance of the company’s data slurper component.

Once your data is uploaded, digiMine cleans it, purges it and breaks it down by means of its data-mining algorithms.

Most customers will probably choose to refresh their data and mine it at night, so that first thing in the morning they can read the reports digiMine compiles.

When we wanted to view our reports, all we needed to do was log into digiMine’s Web site – using either clear text or, preferably, secure socket layer (SSL) for extra security – and view them through a Web browser. We could access our reports from anywhere, so long as we had access to a Web browser and an Internet connection. This capability makes digiMine a compelling addition to sales and marketing meetings.

DigiMine’s reporting capabilities impressed us, and we think they far outstep many of the (considerably less expensive) slick-stream analysis solutions marketed by vendors such as WebTrends and WebSideStory.

To be sure, we could use digiMine to view information concerning site traffic, hourly activity and page views per visit, but it also provided us with a variety of considerably more esoteric reporting features that dealt with such issues as category analysis, category affinity and user segmentation – each of which can be determined only by means of data mining, by the way. DigiMine’s “funnel” reporting feature, which can help sales and marketing executives ascertain the exact point at which they’re losing potential Web sales, could be the capability that makes or breaks an e-commerce start-up.

DigiMine also includes a facility that let us export out Web-based reports to a spreadsheet program such as Microsoft Excel – an important touch. Many executives and most knowledge workers live inside programs such as Excel. The ability to download reports and easily call them up in spreadsheet format is not only a godsend, it’s also a necessity.

By: Stephen Swoyer

Source: Planet IT


Leave a Reply