--- title: Weeplaces keywords: fastai sidebar: home_sidebar summary: "Implementation of Weeplaces dataset." description: "Implementation of Weeplaces dataset." nb_path: "nbs/datasets/datasets.weeplaces.ipynb" ---
{% raw %}
{% endraw %} {% raw %}
{% endraw %} {% raw %}

class WeeplacesDataset[source]

WeeplacesDataset(*args, **kwds) :: Dataset

Dataset base class

{% endraw %} {% raw %}
{% endraw %} {% raw %}
root = './data'

# Define train/val/test datasets on user interactions.
train_dataset = WeeplacesDataset(root, is_group=False, datatype='train')  # train dataset for user-item interactions.
n_users, n_items = train_dataset.n_users, train_dataset.n_items
val_dataset = WeeplacesDataset(root, is_group=False, datatype='val', n_items=n_items)
test_dataset = WeeplacesDataset(root, is_group=False, datatype='test', n_items=n_items)

# Define train/val/test datasets on group and user interactions.
train_group_dataset = WeeplacesDataset(root, is_group=True, datatype='train', negs_per_group=5, n_items=n_items)
padding_idx = train_group_dataset.padding_idx
val_group_dataset = WeeplacesDataset(root, is_group=True, datatype='val', n_items=n_items, padding_idx=padding_idx)
test_group_dataset = WeeplacesDataset(root, is_group=True, datatype='test', n_items=n_items, padding_idx=padding_idx)
Downloading https://github.com/RecoHut-Datasets/weeplaces/raw/v2/data.zip
Extracting ./data/raw/data.zip
# train users 6050 # items 25081
# training groups: 15913, # max train group size: 22
{% endraw %} {% raw %}
!tree --du -h -C ./data
./data
└── [9.1M]  raw
    ├── [2.5M]  data.zip
    ├── [794K]  group_users.csv
    ├── [156K]  test_gi.csv
    ├── [433K]  test_ui_te.csv
    ├── [635K]  test_ui_tr.csv
    ├── [498K]  train_gi.csv
    ├── [3.5M]  train_ui.csv
    ├── [ 72K]  val_gi.csv
    ├── [205K]  val_ui_te.csv
    └── [300K]  val_ui_tr.csv

 9.1M used in 1 directory, 10 files
{% endraw %}