---
license: mit
language:
- en
tags:
- nlp
- ml
- dataset
- fake-news
- classification
pretty_name: x_g85_fn_dataset
configs:
- config_name: processed
  data_files:
  - split: train
    path: fn_train.csv
  - split: test
    path: fn_test.csv
  - split: valid
    path: fn_valid.csv 
dataset_info:
  features:
    - name: text
      dtype: string
    - name: label
      dtype: int32
---


# X_G85 Fake News Dataset


It is a preprocessed dataset that is used to build X_G85 ML Models. The collection of fake news that are collect from the following datasets

## How to stream dataset & use as pandas dataframe

By streaming the dataset, it won't download on your host computer. Read more here [hugging face streaming dataset](https://huggingface.co/docs/datasets/stream).

```py
import pandas as pd
from datasets import load_dataset
```

> Note: The following operation may take some time depending on the size of the dataset.

```py
dataset = load_dataset("x-g85/x_g85_fn_dataset", streaming=True)

train = pd.DataFrame(dataset["train"])
valid = pd.DataFrame(dataset["valid"])
test = pd.DataFrame(dataset["test"])
```

```py
X_train = train["text"]
y_train = train["label"]

X_valid = valid["text"]
y_vaild = valid["label"]

X_test = test["text"]
y_test = test["label"]
```

## Credit

We have used the following datasets to create our own datasets and train models.

- [Kaggle: Fake news detection dataset english](https://www.kaggle.com/datasets/sadikaljarif/fake-news-detection-dataset-english)
- [Kaggle: Liar Preprocessed](https://www.kaggle.com/datasets/khandalaryan/liar-preprocessed-dataset)
- [Kaggle: Stocknews](https://www.kaggle.com/datasets/aaron7sun/stocknews)