HyperTab
HyperTab is a hypernetwork-based classifier for small tabular datasets.
It's especially efficient when the number of samples is smaller than 500. The smaller the dataset, the larger is the advantage of HyperTab over other algorithms.
Installation
pip install hypertab
Usage
from hypertab import HyperTabClassifier
DEVICE = "cuda:0" if torch.cuda.is_available() else "cpu"
clf = HyperTabClassifier(0.5, device=DEVICE, test_nodes=100, epochs=10, hidden_dims=5)
clf.fit(X, y)
clf.predict(X)
Performance
| Dataset | XGBoost | DN | RF | HyperTab | Node |
|---|---|---|---|---|---|
| Wisconsin Breast Cancer | 93.85 | 95.58 | 95.96 | 97.58 | 96.19 |
| Connectionist Bench | 83.52 | 79.02 | 83.50 | 87.09 | 85.61 |
| Dermatology | 96.05 | 97.80 | 97.21 | 97.82 | 97.99 |
| Glass | 94.74 | 46.96 | 97.02 | 98.36 | 44.90 |
| Promoter | 81.88 | 78.91 | 85.94 | 89.06 | 83.75 |
| Ionosphere | 90.67 | 93.43 | 92.43 | 94.52 | 91.03 |
| Libras | 74.38 | 81.54 | 77.42 | 85.22 | 82.72 |
| Lymphography | 85.94 | 85.74 | 87.19 | 83.90 | 83.93 |
| Parkinsons | 86.35 | 74.96 | 86.84 | 95.27 | 80.20 |
| Zoo | 92.86 | 72.62 | 92.62 | 95.27 | 89.05 |
| Hill-Valley without noise | 65.53 | 56.39 | 57.33 | 70.59 | 52.71 |
| Hill-Valley with noise | 58.45 | 56.06 | 55.66 | 67.56 | 51.09 |
| OpenML 1086 | 60.61 | 33.33 | 51.24 | 76.60 | 68.39 |
| Heart | 79.17 | 82.62 | 81.10 | 83.33 | 82.38 |
| Mean rank | 3.50 | 3.78 | 3.07 | 1.35 | 3.29 |
1 maintainer
Metadata
1 maintainer
Release files for hypertab 0.6.2
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| hypertab-0.6.2.tar.gz | 11.0 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| hypertab-0.6.2-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 22.9 kB
Release files / hypertab-0.6.2.tar.gz
| Download URL | hypertab-0.6.2.tar.gz |
|---|---|
| Size | 11.0 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
2a428a0db93488ca4ff3572b076af6a8a1eb685375cbf6a4cffe49addc25965d
|
|
BLAKE2b-256 checksum How to use checksums |
5dd6f34c80f5c02f801106ab4ae248246c0b36052bb3b3b742e2aef51d0d7fe3
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/4.0.2 CPython/3.9.16
|
Release files / hypertab-0.6.2-py3-none-any.whl
| Download URL | hypertab-0.6.2-py3-none-any.whl |
|---|---|
| Size | 11.9 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
1a58fb9c9c5e8f98aee6b1e3ea29fc822467878b2ea54d3e71a934f99b71fb4b
|
|
BLAKE2b-256 checksum How to use checksums |
a546cc4ca7c63c567fa993ae267737433b356c820843c84e5917090dff521403
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/4.0.2 CPython/3.9.16
|
