Week 2/27

[agallagher23]

Completed Tasks:

• Rescheduled midterm presentation meeting with client
• Prepared for midterm presentation
• Practiced midterm presentation with group

[reinabhatkuly]

Completed Tasks:

• Researched SHAP plots and Shapely Values
• Completed Midterm Presentation Slides
• Prepared for midterm presentation with group

[everettbu]

Completed Tasks:

• Prepared for midterm presentation with group
• Made edits to midterm presentation slides
• Assigned models for upcoming week

[xybljy0122]

Completed Tasks:

• Prepared for midterm presentation with group
• Made edits to midterm presentation slides
• Prepared for midterm presentation with group

[xybljy0122]

Notes: 4 graphs presented in the slides are post hoc analysis, can be applied to all model (interpretable AND blackbox), post hoc are approximations Interpretable model can also be explained by inherited predictability. We can manually calculate the exact value.

Suggestions on specific graphs: use PiML to plot main effect of the explainable model compare with post hoc analysis(pdp ice etc), and see how much information are missing

good candidate for blackbox model: look at accuracy results(?), model diagnostics