Rapid-I Help?

blurngr

Hello all!

I stumbled across rapid-i a few months back, and I'm extremely, extremely impressed.  I have a very limited background in statistics / data mining, but we have two upcoming projects that could potentially really benefit from this type of product.  Both of our upcoming projects require looking at time series data.

We are looking for some help with the various aspects of rapid-I and data mining, occasionally, over the next few months.  Probably would require some help transforming time series data to be used by svm, finding outliers in time series, and integrating Rapid-I into our Java application.

It does appear that rapid-I has some consulting services ..... but some of the work we are doing may be "exploratory" -- something more fit for (perhaps) some research work, minor consulting, or a masters thesis of some kind.  I've kind of run to my wits end looking for talented data miners ..... but I'm sure there are some here!  Anyone here able to lend a hand?

I'm happy to send more details about what we're trying to do, if you are at all interested!

Thanks!

--
Anthony

land

Hi Anthony,
you probably meant RapidMiner help, correct?
Although this seems not to be adressed at me since I work for rapid-i, I'm quite curious what you are planing to do. If you could give some details, we would at leaste be able to give some hints.

Greetings,
  Sebastian

blurngr

Right, I did mean RapidMiner.  I assume that's the product name, and rapid-I is the company?  And while you do work for the "company", I suppose these issues may be something that I could contact you about, as it appears that the company behind RapidMiner does consulting work.

What we are trying to do (right now) is somewhat complex time-series forecasting.  For example, inventory or sales forecasting.  I don't think this is anything new .... but what IS new is the vertical market we are trying to pursue.

A more specific example:  We have sales data (both gross sales and item-level sales), and we have some other temporal information, such as the date /time of the sale, whether or not it was a week day / weekend, the month, the weather for the day, etc...  Using this information, we'd like to be able to forecast the next N days - in an "automated" fashion.

In addition to this type of forecasting, we'd like to use the same data set to look for outlying data -- in a method that we can then publish this information back to allow managers to further investigate these outliers.  Again, this is probably not anything "new".

Our team is not as strong in data mining as I'd like them to be, so tackling the basics (like getting time series information into a format that SVM can handle) is proving somewhat cumbersome.

Thanks for your questions!