Charting and Data Manipulation.

Question

Hi, Although there are some threads on here about .net charting controls, I'm starting new thread becuase I'm possibly looking for some advanced data manipulation (maybe this would fall under datamining but I'm not sure) along with charting.

I've been asked to research and prototype and Key Performance Indicators (KPI) system. Basically we have a load of sales figures in a database and we'd like to graph/chart various relationships in this data. For example we increased prices in January 2008, compare the sales for each month with the sales from the prevoius year and display in a chart.

The database doesn't exist yet - we are going to pull the data from various other databases, so the new database would be created from scratch. At the moment I have some example data in excel format and a kind of a prototype created in excel as well.

I'm thinking there are probably two ways to approach this:

1. Manually create all the queries I need and supply the data to the .net charting control.

Or

1. Is there some off the shelf data manipulation / mining tool that could do a lot of the grunt work in pulling out relationships in the data. Not sure if this kind of thing exists? (though I know I attempted to build something like this for my dissertation - it was crap beyound belief!)

One a related note, do Dundas charting and microsoft asp.net charting tools do essential the same stuff. In a thread on stack overflow someone claims there are actually the same thing, did MS buy dundas? Is there any reason to buy dundas rather than use the free MS chart control?

Any advice on this would be greatly appreciated. Thanks in advance!

Answer 1

Microsoft use quite a bit of Dundas Technologies in their products. The graphs in SQL Server reporting services for instance are from Dundas, as are the graph controls in the ASP.NET Charts. You can read more about the controls and using them effectively in this blog from a blogger on the SSRS team

Answer 2

Sounds like a job for a standard statistical package, such as SAS, SPSS-PASW (formerly, Clementine), R, Stata, or S-Plus.
While it is useful (and more scientifically sound) to have some kind of framework structuring your data mining approach, you can implement routines in these programs while remaining agnostic about what you expect to find (or how you expect to find it):

http://www.sas.com/technologies/analytics/datamining/

You can import or query your data into any of these programs from your DB, and most of these programs include routines to let you query a lot of DBs directly (e.g., SQL, mySQL, PostGreSQL) and therefore enabling you to automate the process of extracting, data mining, and visualizing/graphing your data with impunity.