# New to Data Mining, which algorithm to use for my purpose?

### Question

• Hi,

I'm, as you can see, new to Data Mining and need som help to choose the right algorithm for my case.

My case is:

We want to predict the probability that an answere in a surevey is correct.
In the survey, lets say a question could be: How much power did your house use last month?

If the persons taking the survey answere 100kw, but the norm for a house of that size with this amount of house members is 2000kw i want to generate a kind of message to the user.

I know some of this must be done with programming, but this is going to be handles by somebody else than me. My task is to make the model. What I want to know is which algorithm is the best for me to use? The way I see it, the algorithm must do this in a best possible way:

Calculate the norm based on all questions and answeres in the database, take in parameters like year, type (of huse), size of members and the answere for the question and put out the probability of how right the answere is likely to be based on the norm.

I appreciate all the help I can get :)

(And sorry my some poor language, I'm not that use to write english as I am from Norway :-p)

Thanks

Sunday, March 28, 2010 10:24 AM

• Calculate the norm based on all questions and answeres in the database, take in parameters like year, type (of huse), size of members and the answere for the question and put out the probability of how right the answere is likely to be based on the norm.

How many combinations on year , type and size of members do you have ? If too much it can be hard to calculate norm of answers . In this case there isn't much data mining.

If you don't want to calculate so much norms , you can try to use Microsoft_Clustering

Wednesday, March 31, 2010 12:41 PM

### All replies

• Calculate the norm based on all questions and answeres in the database, take in parameters like year, type (of huse), size of members and the answere for the question and put out the probability of how right the answere is likely to be based on the norm.

How many combinations on year , type and size of members do you have ? If too much it can be hard to calculate norm of answers . In this case there isn't much data mining.

If you don't want to calculate so much norms , you can try to use Microsoft_Clustering

Wednesday, March 31, 2010 12:41 PM
• Calculate the norm based on all questions and answeres in the database, take in parameters like year, type (of huse), size of members and the answere for the question and put out the probability of how right the answere is likely to be based on the norm.

How many combinations on year , type and size of members do you have ? If too much it can be hard to calculate norm of answers . In this case there isn't much data mining.

If you don't want to calculate so much norms , you can try to use Microsoft_Clustering

Wednesday, March 31, 2010 12:41 PM
• Calculate the norm based on all questions and answeres in the database, take in parameters like year, type (of huse), size of members and the answere for the question and put out the probability of how right the answere is likely to be based on the norm.

How many combinations on year , type and size of members do you have ? If too much it can be hard to calculate norm of answers . In this case there isn't much data mining.

If you don't want to calculate so much norms , you can try to use Microsoft_Clustering

Wednesday, March 31, 2010 12:41 PM