Tutorial 1C: Machine Learning Interpretability Tutorial

In this tutorial, we will be working with a default of credit card clients dataset in order to understand and be able to interpret the results from Driverless AI. You will also explore how to launch an experiment, create ML Interpretability report, explore explainability concepts such as Global Shapley, partial dependence plot, decision tree surrogate, K-LIME, Local Shapley, LOCO and individual conditional expectations. 

Key:

Complete
Failed
Available
Locked
Objective
Select the "Read" button to begin.
Select the "Read" button to begin. Explore the goal of this tutorial
Prerequisites
Select the "Read" button to begin.
Select the "Read" button to begin. Find general information about what you need in order to complete this tutorial
Task 1: Launch Experiment and MLI
Select the "Read" button to begin.
Select the "Read" button to begin. Download the dataset and upload it to your lab environment. View the details of your dataset, choose your target column, adjust the experiment settings and launch the experiment.
Task 2: Industry Context and ML Explainability Concepts
Select the "Read" button to begin.
Select the "Read" button to begin. Review general concepts about Machine Learning Explainability such as Response Function Complexity, Scope, Application Domain and others.
Task 3: Global Shapley Values and Feature Importance
Select the "Read" button to begin.
Select the "Read" button to begin. Review important concepts about Global Shapley Values, and Feature Importance; also, check out their respectively plots.
Task 4: Partial Dependence Plot
Select the "Read" button to begin.
Select the "Read" button to begin. Review the concept of Partial Dependence, as well as the Partial Dependence Plot
Task 5: Decision Tree Surrogate
Select the "Read" button to begin.
Select the "Read" button to begin. Explore the concept about Decision Tree Surrogate, as well as the Decision Tree Surrogate Model used in Driverless AI
Task 6: K-LIME
Select the "Read" button to begin.
Select the "Read" button to begin. Review the K-Lime concepts, as well as the K-Lime plot and its Advance Features
Task 7: Local Shapley and LOCO
Select the "Read" button to begin.
Select the "Read" button to begin. Explore Local Shapley and LOCO, and see how they can be used in Driverless AI
Task 8: Putting it All Together and ICE
Select the "Read" button to begin.
Select the "Read" button to begin. Take a look at the Dashboard View and ICE technique
Next Steps
Select the "Read" button to begin.
Select the "Read" button to begin. Check out the next tutorial about Time Series - Retails Sales Forecasting
Quiz
25 Questions  |  2 attempts  |  20/25 points to pass
25 Questions  |  2 attempts  |  20/25 points to pass
ML Interpretability Badge
  |  Badge available
  |  Badge available After passing the quiz, you can get your badge by clicking on "Badge Earned". Please check your email address for instructions on how to view, manage, and share your new badge!
You must be logged in to post to the discussion
  • RS

    Could it be possible to at least AUGMENT the tests attempts to 5 as in other cases ? Thanks. Same for the SECOND part of the Module 6 (Tutorial 5A: Disparate Impact Analysis Tutorial).

    Reply
  • RS

    Please review the wording of the questions: I checked EVERY single questions and, while written down, I scored only:

    You have 1 out of 2 allowed attempts remaining.Your previous attempt scored 18/25 and did not pass.

    Please adivice...

    Reply
  • SP

    The wording has been review. Thank you Rino. 

    Reply
  • RS

    Is there a typo in this question ?

    18. Shapley explanations are a technique with credible theoretical support that presents globally consistent global and locally accurate variable contributions.

    Reply
  • SP

    The typo has been fixed. Thank you Rino. 

    Reply
  • RS

    In the following paragraph:

    9. Task 7: Local Shapley and LOCO

    first block is not clear; maybe a typo (in bold - maybe the phrase should sound "that presents consistent global and locally accurate variable contributions"):

    Shapley explanations are a technique with credible theoretical support that presents globally consistent global and locally accurate variable contributions. Local numeric Shapley values are calculated by repeatedly tracing single rows of data through a trained tree ensemble and aggregating the contribution of each input variable as the row of data moves through the trained ensemble.

    Reply
  • SP

    Your feedback will be incorporated soon. Thank you. 

    Reply
  • SP

    Thank you for the feedback. 

    Reply
  • RS

    Which of the LABS in Aquarium is the one for this Tutorial ? Please state it clearly (I suppose 

    Driverless AI Training (1.9.0)

    Lab ID:4Lab duration:120 minutes

    )

    Reply
  • FV

    Rino, these tutorials are meant to be done in the lab called "Driverless AI Test Drive" as specified in the Prerequisite section. This is lab 1, although we do not specify the Lab number, we do specify the name of the lab.

    Reply
  • RS

    The data set comes from the UCI Machine Learning Repository Irvine, CA: University of California, School of Information and Computer Science

    WRONG LINK: hhttps://archive.ics.uci.edu/ml/datasets/default+of+credit+card+clients#

    Reply
  • FV

    Thank you, Rino. we will fix this.

    Reply
  • SC

    After 2 Failed QUIZ attempt , how can take next attempt after reading all the topics. ? Is there any way to re take the quiz after 2 attempt 

    Reply