relational-datasets

A small library for loading and downloading relational datasets.

Use Case 1: Managing Zipfiles of Data

Running the fetch method downloads a version of a datset to your local cache:

import relational_datasets

relational_datasets.fetch("toy_cancer")
relational_datasets.fetch("toy_father", "v0.0.2")
relational_datasets.fetch("webkb")

Resulting in:

~/relational_datasets/
├── toy_cancer_v0.0.3.zip   <--- latest
├── toy_father_v0.0.2.zip   <--- specific version
└── webkb_v0.0.3.zip        <--- latest

Use Case 2: Loading Data

The load method returns train and test folds—each with pos, neg, and facts.

For example: "Load fold-2 of webkb"

from relational_datsets import load

train, test = load("webkb", fold=2)

print(len(train.facts))
# 1344 facts in fold-2 of webkb

Install

From PyPi

pip install relational-datasets

From GitHub Source

git clone git@github.com:hayesall/relational-datasets.git
cd relational-datasets
pip install -e .

Function Signatures: Quick Reference

Load or fetch data:

load(name: str, version: str = "", fold: int = 1) -> Tuple[RelationalDataset, RelationalDataset]
fetch(name: str, version: str = "") -> str

Get or clear the cache directory:

get_data_home(data_home=None) -> str
clear_data_home(data_home=None) -> None

Helper method to check the latest version of datasets on GitHub:

latest_version() -> str