Skip to content

Dataset · JanosAudran

SEC financial reports with sentiment

JanosAudran/financial-reports-sec

US companies 10-K annual reports, split into sections and sentences, with sentiment labels.

What's inside

The dataset contains annual reports of US public companies from the SEC EDGAR system. Each 10-K report is split into 20 sections, and sections into individual sentences. Each report has a sentiment label based on market reaction around the filing date and additional metadata. Language is English.

What you can do with it

  • Training text classification for financial texts
  • fill-mask on financial reports
  • Sentiment analysis of SEC reports
Tasks
Fill mask, Text classification
Languages
English
Size
10M–100M rows
Downloads this month
764
Likes
78

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("JanosAudran/financial-reports-sec")