Looping in load script (KNN Algorithm)

[#1]

Looping in load script

I have the following data:

Customerid	Age	Income (k)	Purchased
1	45	46	Book
2	39	100	TV
3	35	38	DVD
4	69	150	Car Cover
5	58	51	CD player

What I like to do is to find the nearest neighbor,

Meaning, which customer id are relatively close to each other.

The formula I’m using is:

SQRT(((( (customerid(X)Age) - customerid(Y)Age))/(MAX(age)-Min(Age)))^2) + (((customerid(X)Income) - (customerid(Y)Income))/(MAX(Income)-MIN(Income)))^2 )

What I like to do is, run this in a loop in the load script,

And get the nearest neighbor for each customer id.

My expected output should be:

Customer, neighbored, score

For example:

For customer 5

SQRT((((58 - 45)/(69-35))^2) + ((51 - 46)/(150-38))^2 ) = 0.38495

Customer =5

Neighbor = 1

Score = 0.38495

Checking customer 5 against other customers will result a higher scores, so eventually, I need the minimum for each customer that was checked.

Thanks for your help,

Tomer

[Gysbert_Wassenaar]

See attached qvw. I don't think this is suitable for very large number of records. You may want to use a specialized tool voor this kind of analysis. Maybe R with the Rweka package.

[#3]

thanks its very helpful.

I will give it a try with Weka as well.

Cheers,

Tomer