> ML_LIBRARY // CAUSALML_v1.0
CausalML
Uber Open Source — Uber's production uplift modeling and causal machine learning library for targeted campaigns.
probabilistic-modellingv0.15.2Apache-2.0qualified
Model Training
Accelerators:
CPU
Distributed Training:No
Model Inference
Inference Accelerators:
CPU
Deployment Targets:server
What It Does
- +Uplift modeling and meta-learners (S-Learner, T-Learner, X-Learner, R-Learner) wrapping LightGBM and XGBoost
- +Tree-based uplift algorithms (Uplift Random Forest, Uplift Tree)
- +Uplift evaluation metrics: Cumulative Gain, Qini curve, and Area Under Uplift Curve (AUUC)
- +Interpretable causal tree visualizations
What It Does Not Do
- -Perform automated causal discovery without intervention logs
- -Run real-time on edge microcontrollers
- -Train deep vision transformer architectures
>Suitable Work Types
- Targeting marketing promotions exclusively to "persuadable" customers who only convert with a discount
- Minimizing ad spend cannibalization by excluding "sure things" who convert regardless
- Evaluating A/B experiment uplift curves across user segments
>Unsuitable Work Types
- Pure natural language or speech generation
- Standard supervised classification where treatment groups do not exist
Data Residency Implications
Runs entirely in local memory on internal servers. Zero telemetry.
Security Considerations
Apache-2.0 license with zero commercial restrictions.
Operational Profile & Known Limitations
Maturity:mature
Learning Curve:moderate
Ops Complexity:low
Cost Tier:free-oss
> Known Limitations:
- Requires randomized control trial (RCT) treatment flags or carefully controlled propensity scores to avoid biased uplift estimates.
Associated Incident Patterns (Incidentpedia)
Enforce safeguards and monitoring to guard against these documented real-world failure modes:
> Primary Evidence & Benchmark Citations
CausalML Documentationofficial-docs • >=0.14.0, <=0.15.x
2026-09-25HIGH