Analyse Food ?

Question

Dear guys,

I'm working on a project called "healthy food project". It is in the early birth stage, not even birth yet.  Data are in html format and I wonder how should I go about Data mining it so it creates meaningful results. The data can also be in RDBMS (which are basically stored in a MySQL database).

What is your advice to first start this ?

I'm just a student so I can't afford the training. I just need some other suggestion other than the basic tutorials which of course I find really useful. Since it is a Data mining forum, about ideas.

Thank you.

Iced.

iced · Answer

wessel  wrote:
Can you translate this task into a prototypical task studied in literature?

You wish to predict to which group people belong based on their click pattern on your website?

Hi wessel, you're right I should put this down into a better prototypical written statements. Give me some time. Thanks for reading this post!

wessel · Answer

Can you translate this task into a prototypical task studied in literature?

You wish to predict to which group people belong based on their click pattern on your website?

iced · Answer

Hi Ingo,

Thanks, I've developed a Web portal. This portal collects data such as Age, Gender and Location automatically and indirectly about the users. There are other data, but at the moment keeping to this boundary for experimentation's sake.

I'm going to analyse something out of the groups of users. I intended to use rapidminer as of today since my university's SAS version has expired and I am graduating and leaving them soon, might get a job at SAS though, but at the moment just focusing on something that doesn't expire.

I was able to generate a simple analysis grouped by group_id and number of users in each group in SAS. If I were to use rapidminer, I'm sure it can be done, but I'm in the process of expanding the analysis and moving on to the mining part. The approach of this project adopts hybrid collaborative filtering (CF) and the memory based is taken care of using Slope One algorithm. I'm thinking to use rapidminer as the model based.

The combining effort of Slope One with rapidminer is a bit unclear if it can be done, whichever method, manually, or not so automatically.  But I'd like to hear some opinions here. Thanks.