Online News Popularity with 61 Features from Mashable Articles

OnlineNewsPopularity is a tabular regression dataset from the UCI repository summarizing 61 features for articles published by Mashable over a two-year period. The goal is to predict the number of social media shares, with features including word counts, link counts, and content channel indicators. The dataset was created by researchers from INESC TEC and Universidade do Minho for a 2015 conference on artificial intelligence.

Use Cases

• Predicting article popularity based on textual features like word count and unique token rate.
• Analyzing the relationship between multimedia content (images, videos) and social media engagement.
• Modeling the impact of article metadata, such as keyword attributes, on share counts.
• Benchmarking regression algorithms on datasets with mixed categorical and numerical features.

Strengths

• Contains 61 attributes per record, with 58 designated as predictive features.
• Data is derived from a real-world source, Mashable, over a two-year publication period.
• Includes a variety of feature types such as textual statistics, link counts, and content channel indicators.

Limitations

• Row count is unknown, which may limit suitability assessment for large-scale modeling.
• The original article content is not included, only derived statistics.
• Last update date is unknown; freshness unverified.

Provenance

Source

UCI Machine Learning Repository

Collection Method

Features were summarized from articles published by Mashable.

Time Range

Articles published over a two-year period prior to acquisition in January 2015.