"Weight by Value Average" has reversed output weights

mdc

Hi,

I tried to use the "Weight by Value AVerage" operator for text mining but noticed that the characteristic features received the lower weights rather than the higher weights. It is the opposite of what is described in the description. Is it a bug?

Matthew

land

Hi,
this depends on your data? Do the characteristic attributes really have the higher values in average? You might check it, by filtering out examples of all other classes and take a look at the meta data view's average column.

Greetings,
  Sebastian