"Advice to get started. Linking an SQL and Excel database"

Pepijn

Dear all,

I like to get some advice on how to get started.
This is my situation:

I work in a small company where we make foam. We have a production hall where the data of all the machines is stored in an SQL database. We have a laboratory where all the data of physical testings are stored in an Excel sheet.
Now my boss would like to see those two combined to check if there are correlations between the productionprocess and the physical behavior of the foam.

I think this is possible with Rapidminer but I don't know how..jet. The things I want advice in are:
-The online tutorial is simple in a way that I understand what's going on (great), still I lack the insight of when to use what if I get started with an example set of my own. Can somebody advise me for a document with the very basics.
-How can I deal with Excel if:
column 1&2, line 1(merged cell)= test title
column 1 and 2, line 2= two different demensions of test rusult
rest of the lines are test results
-How to combine the SQL and Excel database in the most efficient way

My background is lab-analyst and I did study for chemist half way. At my work I do quality control and R&D projects. The statistical results Rapidminer gives (the examples I've seen) I can interpret.

Thanks

TobiasMalbrecht

Hi,

let my try to answer your questions:

Pepijn wrote:
-The online tutorial is simple in a way that I understand what's going on (great), still I lack the insight of when to use what if I get started with an example set of my own. Can somebody advise me for a document with the very basics.

Well, generally the built-in RapidMiner tutorial is a good way to start learning how to actually use RapidMiner, nevertheless assuming that you have at least some basic knowledge of data, their properties as well as a general idea of data mining approaches and what they are good for. Unfortunately, there is actually no self-learning document, which explains both the basics of data mining and their corresponding application in RapidMiner. But there are of course solutions to that: first, you may read a book on data mining basics (e.g. the book "Data Mining" by Witten & Frank) and then try to use the approaches you have learnt from the book in RapidMiner. This however certainly involves a lot of discipline: there is no free lunch . The second - and certainly better - solution to your problem is: attend one of our training courses and you will get (1) an explaination of widely used, state-of-the art data mining approaches, (2) an indepth insight in how to apply these approaches to data in RapidMiner and - most important - (3) answer to your questions. This also includes an advice on how to use RapidMiner in conjunction with your data, if you bring them along. There is more information on our training courses at http://rapid-i.com/content/view/7/95/lang,en/. Don't get me wrong: I do not want to advertise in the first place, I really do think that taking a training course may save you a lot of time learning things on your own and prevents you from doing things wrong.

Pepijn wrote:
-How can I deal with Excel if: column 1&2, line 1(merged cell)= test title column 1 and 2, line 2= two different demensions of test rusult rest of the lines are test results

Unfortunately, this is not yet possible with RapidMiner. But we are planning to make the [tt]ExcelExampleSource[/tt] operator more flexible which probably includes that you will be able to specify the area in a worksheet which should be loaded as data into RapidMiner. This is already on our todo list, but I do not know yet when we actually will have the time to focus on that.

Pepijn wrote:
-How to combine the SQL and Excel database in the most efficient way

Well, that one is complicated in so far, as there are numerous ways to combine data and it depends crucially on the properties of your data as well as the analysis goal which one of these ways is the most suitable. Hence, without seeing your data it is almost impossible to give a reliable advice. To overcome this, you may either post more details about your in this forum and see if we - or other forum users - can help you then or you may of course talk to us concerning professional support or consulting.

Hope, this was helpful in some way. Otherwise we highly appreciate you would post again or contact us.

Regards,
Tobias

Pepijn

Thank you for your quick and clear reply.

If it was up to me, I would come to the training. However since we are a very young and small company, my boss wants to see results first before investing. So I will go for the book than.

About my question of linking the databases, I get your point. Indeed it's like you said. Let me ask a (i think) more simple question.
Is it possible in Rapidminer to import a selected data-table directly out of an SQL data-base?
Till now I converted the data to an excel table via Crystal Reports.

Thanks

TobiasMalbrecht

Hi,

Pepijn wrote:
Is it possible in Rapidminer to import a selected data-table directly out of an SQL data-base?

yes, this is possible. You can directly import data from an SQL database using the [tt]DatabaseExampleSource[/tt] operator. The easiest way to configure the operator is by executing the wizard which lets you specify all the necessary parameter values in a convenient way.

Hope that helps,
Tobias

Pepijn

Thank you for your advice. I will keep you updated of my progress.