Kmeans - Assign Cluster to new data

[trevorwightman]

Hello,

I am working on a K-means project and need a little help for applying this to new data. Here is the scenario:

I have a historical file with many records. Each record has its original data and has been supplemented with a few attributes and results data. I compiled the result fields to one field which is the weighted score of my record on how well it performed. I will be taking the supplemented attributes and running a k-means to split these records up into 10 different clusters. I analyzed each cluster and found the average weighted score. A higher average weighted score means that group of particular records did better on average overall. Once I get new data in I will not have results data and I will be able to add those supplemental fields. How do I assign these new records to the clusters I have previously assigned? I'd like to score these new records so I can determine which are the "best" records.

[ggruccio · Accepted solution]

Hi @trevorwightman,

I think this is the tool you are looking for.  If I remember correctly you feed the model from the original cluster analysis into one input and the new data into the other input.