Define main parameters for k-nn algorithm

Olus

I would like to apply k-nn to a Trading activity analysis. How do I define the parameters on which the k-nn has to base itself? I would like to cluster my data based on some of the parameters such as: P&L, Size of the trade, holding period 
and put other parameters which are not important: Date / Buy Price / Sell Price, etc 

How should I label the parameters I want k-nn to use and the ones it should not take into account? 
Many thanks for your help.

varunm1

Hello @Olus

If I understanding correctly, you are trying to do clustering on your data. If so, the algorithm you are talking about is K-means clustering and not KNN. This is confusing for many, but KNN is a supervised model and k-means is an unsupervised model (clustering algorithm). One of the important parameters that we vary in K-means is the value of K (number of clusters). This is dependent on multiple factors.

Please see below thread where my good friend @lionelderkrikor explain about this,
https://community.rapidminer.com/discussion/comment/61654#Comment_61654

If you are talking about KNN for classification do let us know, we will provide more information.

Olus

Hi @varunm1
Many thanks for your insightful answer. 
I was indeed referring to the unsupervised k-means process. Will review the link you mentioned. 
I also would like to use knn for the supervised part of my analysis, could you send me the link?
Many thanks for your precious time and help. 

Olus