Uplift-modelling task. Rank the customers in descending order of communication efficiency.
The competition page:
- https://retailhero.ai/c/uplift_modeling/
- https://ods.ai/competitions/x5-retailhero-uplift-modeling/data
The data:
- data/clients.csv — clients information
- data/products.csv — products information
- data/purchases.csv — purchase history before sms advertisement
- data/uplift_train.csv — train set of clients with conversion information
- data/uplift_test.csv — test set to model the uplift
Solutions:
- Baseline solution:
uplift_solution.py - Final solution (top-100):
create_dataset.py,pipeline.ipynb
The libraries needed to run the code are listed in requirements.txt.
Reference materials:
- https://github.com/maks-sh/scikit-uplift
- https://github.com/olegitor/X5.Uplift.public
- https://github.com/nersirion/nersirion-RetailHero.ai-uplift
- https://github.com/rekcahd/retailhero_uplift
Tutorials on uplift-modeling (in Russian):