"Applying Anomaly detection operators to categorical dataset"

cazzi123

Hi,

I am trying to apply the anomaly detection operators to categorical datasets. The only preprocessing that I am doing to dataset is removing missing attributes and duplicates. Should I convert the categorical attributes to numerical before I apply the anomaly detection operators or should they be left as they are?

I am attempting to determine if there is an impact on classification results when outlier are removed.

Please let me know your opinion on this.

Thanks,
C/

fras

There are 14 operators dealing with Anomaly Detection...
But in general: It depends on the measure type. If you choose "Nominal Measures" only nominal attributes are used.
You have to play around. Happy mining !