Normalization of the Features

vijaypshah

Hello,

I am newbie to rapidminer. I am intersted in normalizing the attributes between 0-1 Or by the log scale? Is there any operator that allow us to do this easily?

Thanks,
Vijay

IngoRM

Hi Vijay,

this is of course possible: just use the operator "Normalization" after loading the data. The standard settings for this operator perform a z-transformation which will lead to mean value 0 and standard deviation 1. If you untick the corresponding setting, the operator will normalize the data between min and max (default: 0 and 1). This looks like:

<operator name="Root" class="Process" expanded="yes">
    <operator name="ExampleSetGenerator" class="ExampleSetGenerator">
        <parameter key="target_function"	value="sum"/>
    </operator>
    <operator name="Normalization" class="Normalization">
        <parameter key="z_transform"	value="false"/>
    </operator>
</operator>

If you want to transform only some of the columns you can combine the normalization with the operator "AttributeSubsetPreprocessing".

You can find this and all other operator by using the search option which is available in the "New Operator"-Tab (below the group tree) and in the "New Operator"-Dialog which will be shown after clicking on the icon in the icon bar. The normalization and many other operators are also shown in the online tutorial available in the help menu and in the samples delivered together with RapidMiner.

Cheers,
Ingo

vijaypshah

Thanks for quick  and detailed response! I will look in the New Operator more carefully...

Regards,
Vijay