Datasetpublic

Saika-Zh/financial-qa-data

financial-qa-data This project focuses on the collection, cleaning, and preparation of datasets for natural language processing (NLP) tasks.The data is divided into raw and processed versions: raw: original, uncleaned datasets processed: cleaned and ready-to-use datasets 📂 Directory Structure project/ ├── raw/ # Original datasets │ ├── company_history_news │ ├── company_reports │ ├── investopedia │ └── kaggle │ ├── processed/… See the full description on the dataset page: https://huggingface.co/datasets/Saika-Zh/financial-qa-data.

sourceHugging Faceupdated 8mo agoView on Hugging Face

0likes5downloads

Dataset Card

<!--

—@Author: Zhao
—@Date: 2025-12-17 18:25:20
—@LastEditors: Please set LastEditors
—@LastEditTime: 2025-12-17 21:28:17
—@FilePath: README.md
—@Description:
— -->

financial-qa-data

This project focuses on the collection, cleaning, and preparation of datasets for natural language processing (NLP) tasks. The data is divided into raw and processed versions:

—raw: original, uncleaned datasets
—processed: cleaned and ready-to-use datasets

📂 Directory Structure

project/
├── raw/                # Original datasets
│   ├── company_history_news
│   ├── company_reports
│   ├── investopedia
│   └── kaggle
│
├── processed/          # Cleaned datasets
│   ├── company_history_news
│   ├── company_reports
│   ├── investopedia
│   └── kaggle_spli

1. raw

—company_history_news: Historical news data related to companies
—company_reports: Raw company reports data
—investopedia: Data from Investopedia
—kaggle: Dataset from Kaggle for training purposes

2. processed

—company_history_news: Cleaned historical news data
—company_reports: Cleaned company reports data
—investopedia: Cleaned Investopedia data
—kaggle_split: Kaggle dataset, cleaned and split for use

⚙️ Installation

Clone the repository:

bash

git clone <repository-url>

Install required Python packages:

bash

pip install -r requirements.txt

📑Dataset Details

company_history_news: This dataset includes historical news articles related to different companies, which can be useful for sentiment analysis, event detection, and company-related trend analysis.

company_reports: Contains raw company financial reports, which have been cleaned to remove irrelevant sections, making it ready for analysis.

investopedia: Data scraped from Investopedia, cleaned and structured for easy use in NLP tasks such as text classification and summarization.

kaggle_split: A split dataset from Kaggle, pre-processed and cleaned for easy integration with machine learning models.

🧹Data Cleaning Process

The raw data goes through a series of cleaning steps, including:

Removing irrelevant or noisy information.

Structuring the data into a consistent format.

Handling missing values.

Normalizing text (lowercasing, punctuation removal, etc.).

🚀Usage

To use the processed datasets, simply load the relevant files from the processed directory. For example:

python

import pandas as pd

# Load cleaned company history news data
company_history_news = pd.read_csv('processed/company_history_news/data.csv')

# Process further as needed

🤝Contributing

Feel free to contribute to this project by submitting issues or pull requests. If you wish to add new data sources or improve the cleaning pipeline, your contributions are welcome!

📜License

This project is licensed under the MIT License - see the LICENSE file for details.