jester_rating.py

import pandas as pd
import datasets
from sklearn.model_selection import train_test_split
import numpy as np

_CITATION = "N/A"
_DESCRIPTION = "N/A" 
_HOMEPAGE = "N/A"
_LICENSE = "apache-2.0"


class JesterEmbedding(datasets.GeneratorBasedBuilder):

    VERSION = datasets.Version("0.0.1")

    def _info(self):
        features = datasets.Features({
            "user_id": datasets.Value("int64"),
            "item_id": datasets.Value("int64"),
            "rating": datasets.Value("float64"),
        })
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=features,
            homepage=_HOMEPAGE,
            license=_LICENSE,
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        urls = "./jester_rating.parquet"
        data_dir = dl_manager.download_and_extract(urls)
        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                # These kwargs will be passed to _generate_examples
                gen_kwargs={
                    "filepath": data_dir,
                    "split": "train",
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.VALIDATION,
                # These kwargs will be passed to _generate_examples
                gen_kwargs={
                    "filepath": data_dir,
                    "split": "dev",
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.TEST,
                # These kwargs will be passed to _generate_examples
                gen_kwargs={
                    "filepath": data_dir,
                    "split": "test"
                },
            ),
        ]

    def _generate_examples(self, filepath, split):
        df = pd.read_parquet(filepath)

        rng = np.random.RandomState(42)
        
        train_df, test_df, val_df = [], [], []
        
        for user_id in df.user_id.unique():
            if len(df[df.user_id == user_id]) < 3:
                continue
            _train_df, _test_df = train_test_split(df[df.user_id == user_id], test_size=0.2, random_state=rng)
            _train_df, _val_df = train_test_split(_train_df, test_size=0.2, random_state=rng)
        
            train_df.append(_train_df)
            val_df.append(_val_df)
            test_df.append(_test_df)
        
        train_df = pd.concat(train_df)
        test_df = pd.concat(test_df)
        val_df = pd.concat(val_df)

        if split == "train":
            for _id, row in train_df.iterrows():
                user_id, item_id, rating = row
                user_id, item_id = int(user_id), int(item_id)
                yield _id, {"user_id": user_id, "item_id": item_id, "rating": rating}
        elif split == "test":
            for _id, row in test_df.iterrows():
                user_id, item_id, rating = row
                user_id, item_id = int(user_id), int(item_id)
                yield _id, {"user_id": user_id, "item_id": item_id, "rating": rating}
        else:
            for _id, row in val_df.iterrows():
                user_id, item_id, rating = row
                user_id, item_id = int(user_id), int(item_id)
                yield _id, {"user_id": user_id, "item_id": item_id, "rating": rating}