Comments (2)
The distinction is somewhat subtle - the difference is that only variables in X are used to calculate the heterogeneous treatment effect strength \theta(X). So you will need to decide how you want to partition your confounders between controls W that potentially affect T and Y but not the strength of the treatment effect, and features X that affect T and Y and can also affect the strength of their relationship.
As one example, if you put everything into W instead of X (and just pass None for X), then will end up calculating the average treatment effect rather than a heterogeneous treatment effect. At the other extreme, if you put everything into X and pass None for W, you are saying that the treatment effect can vary depending on every variable.
The more things you put into X instead of W, the more flexible your model is, but the harder the estimation problem becomes, and so you should expect your estimates to be noisier.
If the ultimate goal of your estimation task is to come up with an individualized policy, it may also be the case that there are certain variables that you don't want to consider as policy-relevant (e.g. maybe you are legally prohibited from using, say, gender to generate different treatment policies), in which case it may make sense to leave those in W instead of X even if you think that they may change the effect.
from econml.
Really detailed explain! Thanks!
from econml.
Related Issues (20)
- CausalForestDML with binary outcome and treatment HOT 2
- Binary treatment and Continuous outcome HOT 1
- Misleading warning for multiple treatments HOT 2
- Behavior with Discrete Treatments (discrete_treatment=True) HOT 2
- dowhy is no longer a dependency of the main econml package HOT 4
- IV Models: CATE or LATE? HOT 1
- Background Variable for Causal Analysis Object HOT 2
- Scaling Othrolearners using Ray HOT 6
- Only one CATE value is recorded from EconML (when called from DoWhy) HOT 1
- Calculate ATE when T is continuous in double machine learning HOT 4
- PEP 517 support HOT 3
- Policy using Treatment Classes with DynamicDML HOT 1
- Wrong ATE estimation result, expected positive ATE got negative ATE HOT 2
- CausalForestDML need X to successfully run HOT 2
- how can i save model? HOT 2
- Bug with SHAP library v.0.41 HOT 1
- NumbaDeprecationWarning via shap HOT 1
- Changing covariance logic in DRIV HOT 1
- Will DRPolicyForest control for confounder in Treatments that occured under different seasons? HOT 3
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from econml.