Downloads · 30 days
0
uplift-criteo-kdd-logic
nguyennghia0902/uplift-criteo-kdd-logic
uplift-criteo-kdd-logic is a machine learning model from nguyennghia0902. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as mit.
Mô hình ước lượng CATE (Conditional Average Treatment Effect) trên Criteo Uplift Dataset.
Access
Public
Updated May 16, 2026
Repo size
610 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.csv351 MB · 58%
At a glance
- License
- mit
- Access
- Public
- Created
- May 16, 2026
- Updated
- May 16, 2026
- SHA
- fc6e6bae
Datasets
Model card
From the Hugging Face model README
Uplift Modeling — Criteo KDD Project
Mô hình ước lượng CATE (Conditional Average Treatment Effect) trên Criteo Uplift Dataset.
Models
| Artifact | Description |
|---|---|
models/dr_real.joblib | DR-Learner trained on real treatment |
models/cf_real.joblib | Causal Forest trained on real treatment |
models/x_real.joblib | X-Learner trained on real treatment (Best) |
models/dr_placebo.joblib | DR-Learner placebo baseline |
models/cf_placebo.joblib | Causal Forest placebo baseline |
models/x_placebo.joblib | X-Learner placebo baseline |
models/propensity_model.joblib | Propensity score model |
models/tau_dr_real.npy | Predicted CATE array — DR-Learner |
models/tau_cf_real.npy | Predicted CATE array — Causal Forest |
models/tau_x_real.npy | Predicted CATE array — X-Learner |
Key Results
- X-Learner Uplift@20%: 0.01439 (4× vs average)
- Optimal Budget Ratio: 5% (Causal Forest)
- Uplift ROI @ 20% Budget: 13.65% (vs 2.93% of Random Targeting)
- Profit Gain vs Random @ 20%: +$11,995
- Profit Gain vs Send All @ 20%: +$446,745
Usage
from huggingface_hub import hf_hub_download
import joblib, numpy as np
# Load best model
x_model = joblib.load(hf_hub_download("nguyennghia0902/uplift-criteo-kdd-logic", "models/x_real.joblib"))
# Predict CATE
tau = x_model.effect(X_test_fe)
- License
- mit
- Created
- May 16, 2026
- Updated
- May 16, 2026