 |
Look to the future
All OLAP components can work with flat dimensions. Flat dimensions, a single
table as the data source. The dawn of the OLAP analysis. Developers' algorithms
far from perfection, no analytical experience, mid of the nineties. Past century.
But the time goes forward. The Star schema: a fact table, several dimension
tables. De facto, it is a standard for those who does not need more. But the step
forward was made, and the question was asked: Is it possible to use dimension tables
with the hierarchy structure during the OLAP analysis? Except for "Yes", OLAP developers
had no other answers. In that way, hierarchy dimensions appeared and with them -
new long-expected opportunities for the OLAP analysts.
The next step, no less logical, is the snowflake schema in the data warehouse.
It is logically to band together the fields Year, Month and Day, or Products and
Products Categories. For the OLAP developers these are the multilevel dimensions.
Fortunately, RadarCube involves all aforesaid hierarchy types taking into consideration
their differences. Moreover, RadarCube allows working with them not only separately
but in complex making it possible unprecedented: building the multilevel dimensions,
each level of which can contain its own hierarchy. If someone will ask you: Why
so complicated? Just answer: It's not complicated but naturally. There are 5000
members in your hierarchy, then why so much? It is better to group 4980 not relevant
members in the group More, and then your report will become clearer to your investors.
And to place the group More next to the rest of the 20 members - it is so natural,
right?
Nowadays, RadarCube makes the things nobody is able to replicate. But tomorrow multidimensional
dimensions will come into life of the OLAP analysts as multilevel and hierarchy
dimensions have already came. Because it is so naturally. And to talk seriously,
we are not doing impossible but just keep striding along with time! Come with us!
Key features: Cube definition
- Support of the following data sources:
-
A single table or query of any database (any TDataSet successor, no BDE required);
- Several tables or queries of any database organized into the "star" or "snowflake"
schema (any TDataSet successors, no BDE required);
-
Support a set of aggregation functions, such as Sum, Count, Distinct Count, Min,
Max, biased or unbiased variance, median and so on. Supports custom aggregation
functions.
-
Support of numbers, strings and dates as measure values.
-
Support of measure grouping.
-
Support of all cube structure elements of Enterprise OLAP servers (dimensions, hierarchies,
hierarchy levels and their relations). You can build the cube in the Desktop OLAP,
and then, when increasing the size of tables, to port the Desktop cube into the
MS Analysis cube, at that the logical structure of the cube will remain unchanged.
- Support of hierarchies of three types:
-
Parent-child hierarchies;
-
Multilevel hierarchies;
- Mixed from both previous types.
-
An ability to create calculated measures, dimensions, hierarchies and hierarchy
levels based on the data both as the fact table rows and all dimension tables rows
related to it with the foreign keys.
-
An ability to create additional calculated measures, using both the information
from the fact table and values of the aggregated cube cells.
-
Support of calculated members of dimensions or hierarchies.
-
Support of dimension or hierarchy attributes which store additional information
about dimension or hierarchy members.
- Support of supersaturated cubes when one cube cell holds two or more values from
the fact table.
Key features: Data visualization
- Unicode support.
-
An ability to display many measures in the Grid simultaneously.
-
Interactive drill-down support of all types of hierarchies separately for every
grid cell.
-
Support of "Undo" and "Redo" actions during the OLAP analysis.
-
Support of saving and restoring the current OLAP slice.
-
Support of member grouping on any hierarchy level without changing the hierarchy
levels structure.
-
Support of sorting on any level of hierarchy or dimension separately. Ability of
overriding any sorting method.
-
Support of ascending or descending sorting based on any column value in the OLAP
grid.
-
Support of filtering any set of hierarchy members with or without applying these
filters in the OLAP calculations.
-
Support of the auto filtering of the hierarchy members depending on their values
in the grid. An ability for selection some major/minor members, either according
to their rank or on the Pareto principle.
-
An ability to display in the grid not only the values of the measures but a percent
value regarding totals, subtotals, or grand totals. An ability to create your own
context-sensitive rules on measure display (for example, a cumulative sum of measure
values for time dimensions, and just its value for all other cases).
-
The auto image drawing in any grid cell, the auto picture positioning in a cell,
depending on the parameters assigned by a programmer.
-
A complete data output control, a feature for the arbitrary drawing in the grid
cells.
-
Support of the custom drawing on the grid cells.
-
Support of the custom popup menus in the grid.
-
Support of the drag-n-drop of measures or hierarchy members within the grid.
-
Auto wrapping of the cell contents depending on its width. Auto sizing of the column
width with auto wrapping for the very "long" cells of the grid.
-
An ability to edit OLAP data directly in the grid ("writeback" support).
-
Allows to display the current slice as a diagram (any TChart successor). Support
of Drill-down actions in the chart. The source code of the charting component comes
for free.
-
Allows to copy the selected grid area into the clipboard. Support all common office
clipboard formats.
-
Allows to export grid data to many formats: MS Excel (using OLE or directly to XLS-file),
HTML, PDF, WMF, RTF, TXT, BMP, XML. Exports to the RaveReport, ReportBuilder, FastReport
are coming soon. Also supports direct drawing on printer Canvas.
- Easy localization. It is possible to localize the evaluation version as well.
Requirements for the developer
-
Borland Delphi 6, 7, 2005 or 2006 versions.
-
Windows NT/2000/XP/2003
-
GDI+ (for Windows NT/2000 users)
- 256MB memory or more (512MB recommended)
Requirements for the end user
-
Windows 98/ME/NT/2000/XP/2003
-
GDI+ (for Windows 98/NT/2000 users)
- 256MB memory or more (512MB recommended)
Why RadarCube?
While developing the RadarCube architecture, we are trying, apart from the own
experience, to take into consideration not only the Desktop-products but the servers'
solutions, such as Microsoft Analysis 2005 and Oracle OLAP. That is why for those
developers who will chose RadarCube there is no necessity to change a client code
when migrating from the Desktop-version into the MS Analysis platform, and in the
near future - into the Oracle platform. End users will just not see the difference
between the versions!
The events models and the data access ideology in the Desktop version are initially
compatible with the future RadarCube.NET version, that is why if you are planning
to port the own code on the NET Framework platform, then, while transferring RadarCube,
you will not have problems.
RadarCube is a constructor - a big and pretty complex toy. In RadarCube, you
can tune everything. You can use it in the simplest mode of the flat hierarchies
by disabling the features of grouping and filtering, and nothing will differ from
the simplest OLAP components. You can intercept the dialog windows by replacing
them with your own. You can intercept the procedures of the context menu generation,
cell drawing in the grid or the hierarchy panels, use any font, paint them in any
colors, and in general, output the things you need on the screen.
Are you lack of standard functions of aggregation? No problems, write your own
and use them in RadarCube.
If, along with the ordinary dimension members you need additional members, for
example, for intermediate totals, that's not a problem! You can add them. Calculated
measures? Easily! Grouping, sorting, multiple filtering? It's easy too! Do you need
to get a value of any cube cell at any measures in order to later use the data in
your own calculations? A single function will do it for you. At the same time, you
use the same function regardless of the RadarCube version you are working in, and
to which OLAP server you are connecting.
We, the team of the RadarCube developers, are happy to realize that our library
sets new standards for the developers of the OLAP applications unifying Desktop
OLAP and industrial OLAP servers, and giving to the developers of the OLAP applications
new great perspectives! We are glad to stay with you. Good luck!
|
|
Download evaluation version
Download
Desktop OLAP demo
RadarCube.VCL for MS Analysis
HierCube for VCL
License agreement
Version history
 |
|
The groups in RadarCube can be created either programmatically or by end users in
run time. Other hierarchy members can be put into group ether dragging the members
to, or by commands in the context menu
|
 |
|
"Measure show modes" replaced the concept of "sub-functions" in HierCube, offer to
a programmer extremely flexible technique of processing data after aggregation.
For example, the "cumulative sum" mode which you can see on the figure, works only
for definite measures and hierarchies "date-time" in whatever cube slice. It doesn't
depend on sorting or filtering methods. This mode is produced with only 20 lines
of code
|
 |
|
Where it is appropriate, hierarchical levels are tabulated in the grid with the indents.
This allows to spare the useful screen space, and takes users back to the tree-like
hierarchy look they may be used to
|
 |
|
Calculated members is an indispensable tool for those who want to show in the OLAP-report
much more than just aggregated data of the database tables. In the current screenshot
two calculated members are used: "Forecasted 2005" which calculates values of future
sales by the approximation method, and "Degree of trend" which visually demonstrates
sales rates of growth from year to year
|
|
|