Do you know the History of Data Science? READ MORE

Individual Explanations

Mariak
Level 2
Individual Explanations

Hello,

i would to ask few things about how individual explanations on DSS work. Particularly i'm interested in shapley method and i wonder:

1) which explainer is behind the calculations (Kernel, Tree, Deep, Linear etc.),

2) if you use some default values for some parameters, like in order to accelerate the analysis performing  sampling in KernelExplainer e.g. of 1000 rows. 

Thank you!

0 Kudos
2 Replies
tgb417
Neuron
Neuron

@Mariak 

When speaking with my colleague @CoreyS about your question.  He pointed me to this documentation.  Does this being to answer your questions?

From your note it appears that in your question 1 you might be referring to a specific method (Kernel, Tree, Deep, Linear) used in the explainer calculations.  Is that the nature of your question?

On question 2 it looks like the documentation might be describing the answer to your questions.

That's what I've got for now.  Hope this helps a bit.

--Tom
0 Kudos
Mariak
Level 2
Author

@tgb417 

Thanks for the reply, i'm aware of this documentation, and yes in question 1 i was referring to a specific method (Kernel, Tree etc.). As i know Kernel Explainer can be used in any model to compute the explanations. So, i was wondering if dss uses this or depending the model adjusts the explainer (e.g for linear model uses the linear, for tree based uses the tree explainer etc.)

A banner prompting to get Dataiku DSS