requirements.txt

# CrossUserDataset — top-level pinned dependencies
#
# Covers EDA notebooks, preprocessing utilities, and the three-tier benchmark
# (baseline + tabular-NN, DG, DA). Install with:
#
#   pip install -r requirements.txt
#
# Python 3.10 recommended; 3.9–3.11 supported. For a conda environment see
# environment.yml.

# ── Core scientific stack ─────────────────────────────────────────────────────
numpy>=1.24,<3.0
pandas>=2.0,<3.0
scipy>=1.10,<2.0

# ── Visualization (EDA notebooks + figure generation) ─────────────────────────
matplotlib>=3.7,<4.0
seaborn>=0.12,<1.0
plotly>=5.18,<6.0

# ── ML baselines + tabular helpers ────────────────────────────────────────────
scikit-learn>=1.3,<2.0
xgboost>=2.0,<3.0
lightgbm>=4.0,<5.0
imbalanced-learn>=0.11,<1.0

# ── Deep learning (MLP / ResNet / tabular NNs / DG / DA) ──────────────────────
torch>=2.1,<3.0
torchvision>=0.16,<1.0  # needed by some DG implementations

# ── Tabular neural networks ───────────────────────────────────────────────────
pytorch-tabnet>=4.1,<5.0
pytorch-widedeep>=1.4,<2.0
deepctr-torch>=0.2,<1.0

# ── Hyperparameter optimization ───────────────────────────────────────────────
optuna>=3.4,<5.0

# ── Manifold learning / embeddings (psych_affect.ipynb) ───────────────────────
umap-learn>=0.5,<1.0

# ── Notebook runtime (EDA) ────────────────────────────────────────────────────
jupyterlab>=4.0,<5.0
ipykernel>=6.0,<7.0

# ── Reproducibility helpers ───────────────────────────────────────────────────
pyyaml>=6.0,<7.0
tqdm>=4.65,<5.0

# NOTE ON DG / DA BACKBONES
# DG methods follow the DomainBed protocols (https://github.com/facebookresearch/DomainBed).
# DA methods follow the Transfer-Learning-Library (TLL) protocols
# (https://github.com/thuml/Transfer-Learning-Library). These repos are
# integrated under ../domain_adaptation/ and are not installed from PyPI;
# see domain_adaptation/README.md for upstream attribution.