Data profiling

Spiros Denaxas spiros at
Fri Mar 14 09:38:29 GMT 2008

On Fri, Mar 14, 2008 at 8:12 AM, Dave Hodgkinson <davehodg at> wrote:

> Is there anything out there that can "profile" data?
> I've got a lot of key/value pairs and I'd like to analyse what "kinds"
> of data are
> associated with particular keys. Bonus points for breaking down
> composite data into
> its constituents and for coming up with analyses of numerical ranges.

One idea would be clustering. Data cluster analysis will group all the keys
together according to some
notion of 'distance' between their values. Its then up to you to examine the
clusters and see if you can deduct any valuable information from them.


> --
> Dave Hodgkinson
> Back in the Market Risk swamp

More information about the mailing list