The Altair Community is migrating to a new platform to provide a better experience for you. In preparation for the migration, the Altair Community is on read-only mode from October 28 - November 6, 2024. Technical support via cases will continue to work as is. For any urgent requests from Students/Faculty members, please submit the form linked here
Weights as 'Global importance of attributes' - method of calculation?
Hi rapidminers,
Auto model has 'General' tab in results, which in turn has 'Weights' section.
The explanation goes like "Weights: the global importance of each Attribute for the value of the target Attribute, independent of the modeling algorithm."
What is 'global importance' in this case? What exact algorithm is used to calculate those weights specifically in auto-model process?
Tagged:
0
Answers
tagging @IngoRM
Scott
Hi Vladimir,
Auto Model uses the correlation (squared Pearson) between the attributes and the label column. By the way, the next version will actually change the calculation slightly to better cover nominal values and improve understandability:
Cheers,
Ingo
Thanks Ingo,
So the closest operator to mirror this process would be 'Weight by correlation' I guess?
Vladimir
http://whatthefraud.wtf
Yes, that's the operator which is used. I will attach the new, complete processes below as well. Please note that they will be different for numerical labels and nominals ones. But please note that those processes have not been published yet in the current version of AM...
Numerical Labels:
Nominal:
Thanks Ingo, pertty helpful (and impressive, too!)
Vladimir
http://whatthefraud.wtf