Individuals automatically create electronic data through online activities, as well as through their offline activity as the latter is captured electronically and often linked to online identities. Data generated by individuals is frequently unstructured in text, image, or video formats, disseminated through multiple platforms and includes:

• Social media posts, such as opinions or reactions on general-purpose sites such as Twitter, Facebook, or LinkedIn, or business-review sites such as Glassdoor or Yelp
• E-commerce activity that reflects an interest in or the perception of products on sites such as Amazon or Wayfair
• Search engine activity using platforms such as Google or Bing
• Mobile app usage, downloads, and reviews
• Personal data such as messaging traffic

The analysis of social media sentiment has become very popular because it can be applied to individual stocks, industry baskets, or market indices. The most common source is Twitter, followed by various news vendors and blog sites. Supply is competitive, and prices are lower because it is often obtained through increasingly commoditized web scraping. Reliable social media datasets that include blogs, tweets, or videos have typically less than five years of history, given how recently consumers have adopted these tools at scale. Search history, in contrast, is available from 2004.