The "Pre compute" tab allows the user to select the pre-processing options to be applied before clustering.
Figure 5: Pre compute tab
The user has the choice or not to set the diagonal of the similarity matrix with zeros.
Moreover, if the similarity matrix is not Gram, it will be multiplied with its transpose so that it becomes Gram.
The user has the choice or not to normalize the data from the dataset.
When data are not normalized, variables with a very wide range of values will influence the search for states more than other variables. Normalization avoids this scaling effect: all variables have the same weight when searching for states. The risk, by using normalized variables, is to give weight to a variable which in reality brings few information.
Next step button
The "Next step" button validates the pre-processing parameters and moves on to the next step which is the clustering part.