Polina Okuneva, Lead Analyst and Team Lead, Glowbyte. Reliable ML.
The talk is intended for a prepared audience: those who have already solved causal inference tasks and understand their main challenges. The most basic methods (T-learner, S-learner, Class Transformation, Two Dependent Models; highlighted in orange on slide 2) are not discussed in detail: there is a lot of material on them in Russian and English, including the author's talk at the NoML community seminar on uplift modeling. The speaker delves into the mathematics of the "advanced" methods, highlighted in green on slide 2: X-, R-, DR-learner, Domain Adaptation Learner, Doubly Robust Instrumental Variable learner and tree-based algorithms (DDP, CTS). For some of the methods, comparisons with other methods from the original papers are provided.
Google or a one-time email code. No password.