Create conjoint-new-to-world-demand-estimation#2490
Conversation
Documentation that explains how conjoint analysis can be used to estimate demand for new-to-world products
NathanielF
left a comment
NathanielF
left a comment
There was a problem hiding this comment.
This is very useful context and point to a "market" of users the current documentation is not serving well. The notes sound like they come from battle tested experience @kaytilann ! Would you be at all interested in turning these notes into a demonstration notebook for conjoint modelling in the product context?
|
|
||
| ## Who This Guide Is For | ||
|
|
||
| This guide is written for product managers and decision scientists who: |
There was a problem hiding this comment.
This is a super useful perspective of discrete choice that is missing from the documentation. Thanks for pushing this.
|
|
||
| It works poorly for: | ||
|
|
||
| - Features that are entirely new to the market and have no existing analog |
There was a problem hiding this comment.
A good problem to flag.
|
|
||
| A more rigorous approach involves treating the two conjoint datasets as inputs to a nested | ||
| model structure -- conceptually similar to a nested logit -- where one model estimates the | ||
| probability of choosing within a subset of products, and a second model estimates choice |
There was a problem hiding this comment.
You might be interested in this PR on consideration sets. It could be used if the attributes that drove consideration were available for new market entrants and were distinct from the utility predictors.
2 participants
Documentation that explains how conjoint analysis can be used to estimate demand for new-to-world products
Description
Related Issue
Checklist
pre-commit.ci autofixto auto-fix.📚 Documentation preview 📚: https://pymc-marketing--2490.org.readthedocs.build/en/2490/