Benchmarks for different forms of target encoder with 10-fold cross validation.
| data_name | categorical features | n_features | n_samples | is_classification | openml_url |
|---|---|---|---|---|---|
| telco | 16 | 19 | 7043 | True | https://www.openml.org/d/42178 |
| amazon_access | 8 | 8 | 32769 | True | https://www.openml.org/d/4135 |
| kicks | 18 | 32 | 72983 | True | https://www.openml.org/d/41162 |
| dresses_sales | 11 | 12 | 500 | True | https://www.openml.org/d/23381 |
| phishing_websites | 30 | 30 | 11055 | True | https://www.openml.org/d/4534 |
| SpeedDating | 61 | 120 | 8378 | True | https://www.openml.org/d/40536 |
| medical_charges_nominal | 8 | 11 | 163065 | False | https://www.openml.org/d/42559 |
| Bike_Sharing_Demand | 4 | 12 | 17379 | False | https://www.openml.org/d/42712 |
| black_friday | 4 | 9 | 166821 | False | https://www.openml.org/d/41540 |
| colleges | 17 | 49 | 7063 | False | https://www.openml.org/d/42159 |
| KDDCup09_upselling | 38 | 230 | 50000 | True | https://www.openml.org/d/1114 |
| KDDCup09_appetency | 38 | 230 | 50000 | True | https://www.openml.org/d/1111 |
| rl | 14 | 22 | 31406 | True | https://www.openml.org/d/41160 |
- Clone repo:
git clone http://github.com/thomasjpfan/sk_encoder_cv
cd sk_encoder_cv- Create virtualenv and install
sk_encoder_cv
conda create -n sk_encoder_cv python=3.8 # or use venv
conda activate sk_encoder_cv
python setup.py develop- Run single benchmarks:
python benchmark.py single adult- Or run all benchmarks
python benchmark.py allThe results will be written into the results directory.
- Generate README report:
python generate_report.py