README.md

metadata

license: mit
task_categories:
  - text-classification
language:
  - en
tags:
  - finance
  - twitter
  - news
  - crypto
  - stocks
  - tweet
pretty_name: Financial Tweets
size_categories:
  - 100K<n<1M

Financial Tweets

This dataset is a comprehensive collection of all the tweets from my Discord bot that keeps track of financial influencers on Twitter. The data includes a variety of information, such as the tweet and the price of the tickers in that tweet at the time of posting. This dataset can be used for a variety of tasks, such as sentiment analysis and masked language modelling (MLM).

We used this dataset for training our FinTwitBERT model.

Overview

This datasets includes all the following three datasets:

• Crypto: https://huggingface.co/datasets/StephanAkkerman/financial-tweets-crypto
• Stocks (and forex): https://huggingface.co/datasets/StephanAkkerman/financial-tweets-stocks
• Other (tweets without cash tags): https://huggingface.co/datasets/StephanAkkerman/financial-tweets-other

Data Description

The dataset comprises tweets related to financial markets, stocks, and economic discussions.

Dataset Fields

The data fields are as follows:

• timestap: The time the tweet was sent.
• tweet_text: All of the text of the tweet, including quoted tweets (prefixed with >).
• tweet_url: The URL of the tweet.
• tweet_type: The type of tweet, this can be tweet, retweet, or quote tweet.
• price_of_ticker: The price of the tickers mentioned in USD ($).
• change_of_ticker: The 24h price change of the tickers in USD ($).
• tickers_mentioned: All the tickers that are mentioned in the tweet.
• category: What type of category the tweet has, the suffix _images means that the tweet included an image.

Usage

This dataset is ideal for training and evaluating machine learning models for sentiment analysis, especially those focused on understanding market trends and investor sentiment. It can be used for academic research, financial market analysis, and developing AI tools for financial institutions.

Acknowledgments

We extend our heartfelt gratitude to all the authors of the original tweets.

License

This dataset is made available under the MIT license, adhering to the licensing terms of the original datasets.