X-means in Rapid MIner Go

Chemical_eng

I have 2 questions about X-means algorithm in Rapid miner Go : 
1. How does it handle categorical variables? 
2. Does it perform automatic normalization of numerical variables or should I do this beforehand? 

alebo

Dear @Chemical_eng,

1. We use one-hot encoding on categorical columns, as long as the number of values is below 50. Any column with more is removed.
2. Yes, the method we use for normalization is called Z-transformation. You can learn more about it here: link under Parameters / method.

After doing these (and more) preparation steps, we run x-means algorithm. Then the cluster result column (called Groups) is appended back to the original data for easier interpretability.

Regards,
Andras

alebo

Dear @Chemical_eng,

1. We use one-hot encoding on categorical columns, as long as the number of values is below 50. Any column with more is removed.
2. Yes, the method we use for normalization is called Z-transformation. You can learn more about it here: link under Parameters / method.

After doing these (and more) preparation steps, we run x-means algorithm. Then the cluster result column (called Groups) is appended back to the original data for easier interpretability.

Regards,
Andras

Chemical_eng

thanks, that answers my question