# Uplift model similar to PyLift or R's Lift Package?

**URL:** <https://discourse.julialang.org/t/uplift-model-similar-to-pylift-or-rs-lift-package/94268>\
**Category:** Machine Learning\
**Tags:** machine-learning\
**Created:** [February 8, 2023, 10:38am UTC](https://discourse.julialang.org/t/uplift-model-similar-to-pylift-or-rs-lift-package/94268 "2023-02-08T10:38:53Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![Billpete002](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/billpete002/32/35091_2.png) [@Billpete002](https://discourse.julialang.org/u/Billpete002)\
**Post date:** [February 8, 2023, 10:38am UTC](https://discourse.julialang.org/t/uplift-model-similar-to-pylift-or-rs-lift-package/94268/1 "2023-02-08T10:38:53Z")

</div>

Hi,

I am looking for a package that works similarly to [PyLift](https://pylift.readthedocs.io/en/latest/quick-start.html) or R’s [lift](https://cran.r-project.org/web/packages/lift/lift.pdf) package. The idea is that I have two models: Self-service and the other is Sales-Assisted. I used Catboost.jl and examined manually that leads going through our sales-assisted model do better, but I don’t want to send all of the customers to sales-assistance - just the ones with the most impact.

Some text on the idea behind this is found [here](https://odsc.medium.com/the-intuition-behind-uplift-modeling-fb63b02421c8).

Having already loaded Conda/PyCall etc. to run CatBoost.jl I could run PyLift - but I was interested if this was already developed as a Julia solution? I didn’t see a package specifically about this, but it might be nestled within another or named differently.

Thoughts?

---

<div class="post-metadata">

**Author:** ![dscolby](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/dscolby/32/204483_2.png) [@dscolby](https://discourse.julialang.org/u/dscolby)\
**Post date:** [February 14, 2023, 4:36am UTC](https://discourse.julialang.org/t/uplift-model-similar-to-pylift-or-rs-lift-package/94268/2 "2023-02-14T04:36:53Z")

</div>

This looks like uplift modeling is estimating the CATE/ITE using ML methods. If that’s the case, you could look into the S/T/X-Learners in [this](https://dscolby.github.io/CausalELM.jl/stable/examples/#S-Learning) package. Instead of using tree-based learners, it uses extreme learning machines, which have less hyperparameters to tune and also perform well on a lot of tasks. That being said, right now you can get the CATE/uplift but I haven’t implemented the bootstrapping procedure for p-values and confidence intervals yet.

Also, this is still really new and should be on the general registry in a few days.