If ad hoc analysis is the goal, why can’t you do it?
Data and business analysts have been restricted for too long with highly structured, highly optimized databases and SQL. It’s practically impossible to do genuine ad hoc analysis in this environment without becoming really good friends with the IT department. And even then, the delays are (or should be) unacceptable.
You already know that trying to do ad hoc analysis through what are really ad hoc reporting tools is cumbersome, inefficient, time-consuming and frustrating. You’re limited by:
- Predesign: you can only ask questions the database was designed to answer.
- Indexing: reporting is limited to the relationships that IT has designed into the database. If you want to explore a nonstandard relationship—or even more significantly, understand what relationships exist that may not have occurred to you during predesign—it's another IT project.
- Summary data: The level of depth in reporting, and hence in your analysis, is limited to the level of detail data included in the database—which, in many cases, is denormalized, prejoined and summarized. While summarized data may be sufficient for some analyses, you really don’t know what’s missing.
IT mitigates these problems by building data marts that support structured or guided queries. Flexibility is increased somewhat, but your analysis is still limited to "what's in the cube"—the data and design provided by IT. This works for a while, but at some point, you realize that you don't have the data you need to perform true ad hoc analysis. Back to IT you go. It doesn't take long for analysts and IT alike to figure out that this is expensive and time-consuming. And it's not the way to run a profitable, nimble business. But, it's best you can do.
Illuminate's correlation database lets you do free-thinking, question-answer exploration. You don't have to explain to IT in advance what questions you'd like to ask. In fact, you don't even have to know all of the questions, only that the source data exists somewhere in the organization and that it's in the data warehouse or data mart. Every data value and correlation is indexed In the database's value-based storage (VBS) structure—automatically—when the data is loaded. That means you can ask anything. Yes, we mean anything, as soon as the data is loaded. That's fast time-to-analytics!
Now you've got the data available to you, it's time to explore the way you really analyze an issue—incrementally, using our iCorrelate desktop ad hoc query tool. You ask, you analyze. You ask again, you analyze some more. We call these Incremental Queries. They let you drill down, drill up, and drill across until you get the answer you need. You don’t need to know the structure of the data, you just need to think and then ask. You can even save the results as a stored procedure to access using standard business intelligence tools for further analysis and/or distribution.
But what if you don't quite know where to start with your analysis? Using an Associative Query, your analysis can take shape when questions come to mind after you identify all relationships associated with any specific data element. And any level of detail can be included. There are no SQL constructs for creating incremental or associative queries.
Use Your Data Fast
To use Excel, SAS, or SQL tools, you spend a significant amount of time staging, scrubbing, preparing, and massaging the data to get it to a point where it's suitable for use in your algorithmic or analytical model. Flat files from transactional systems, in CSV or text format, have to be prepared and cleaned for use. It's time-consuming and painful.
With illuminate's correlation database, check data quality as it loads and spot null values and quality problems. It automatically generates summary statistics about the data, such as number of unique values, min/max and sum for each column. It's easy to understand and evaluate data before use, and it speeds up the preparation, cleansing, and staging of data for any type of ad hoc analysis.
Improve Your Analysis
Illuminate shortens data preparation and staging time and provides new capabilities for performing true ad hoc analysis for tasks such as segmentation, market basket and regression analysis. Combine Associative and Incremental Queries to hone in on purchase affinity by gender, geography or other criteria—with no spreadsheet filtering and inclusion—until you get your answers.
Enhance Your Algorithms
Illuminate runs existing segmentation algorithms and other advanced queries through any ODBC-compliant front-end tool, or by using iCorrelate (freeing you from the constraints of SQL) you can create and store these to run at any time. And as information needs or data changes, there’s no need to convert, prepare, index and transform data for loading.
Algorithmic models are designed using a repetitive process of testing and evaluation to arrive at the best model. The ability to create and rapidly execute Incremental and Associative Queries with iCorrelate simplifies and accelerates model development, and makes continuous refinement—and ad hoc analysis—practical.
Correlation data warehouses and data marts are built and deployed in a fraction of the time as other options. And once built, analytics are fast and genuinely flexible. We invite you to compare the data mining process using conventional RDBMS technology versus illuminate's correlation database, which:
- Reduces data mapping and discovery time for the evaluation of raw source data. This is particularly valuable in environments wherein new types or sources of data are frequently introduced.
- Reduces the time and complexity of data staging and preparation.
- Provides incremental questioning capability for true ad hoc analysis. Illuminate eliminates the constraints of relational databases and the need to load data into spreadsheets for analysis. It enables you to ask any question of your data with no concern for the underlying structure of the database.
- Drastically shortens time-to-analytics and time-to-answer.
- Enables you to build algorithmic models with direct access to the database and without the limitations of SQL.
|
