CoolFace
Datasetpublic

PulkitSahu/The-AI-Thread-Engagement

AI Thread Engagement Rate Predictor Dataset This dataset contains a real-world, manually collected sample of 14 threads posted on X (formerly Twitter) under this account between September 2024 and January 2025. Despite its small size, it is an authentic dataset with real engagement metrics, making it ideal for small-scale experiments, educational purposes, and exploratory analysis of how post features influence engagement. 📌 Purpose The dataset is designed to… See the full description on the dataset page: https://huggingface.co/datasets/PulkitSahu/The-AI-Thread-Engagement.

sourceHugging Faceapache-2.0updated 2y agoView on Hugging Face
0likes16downloads
Dataset Card

AI Thread Engagement Rate Predictor Dataset

This dataset contains a real-world, manually collected sample of 14 threads posted on X (formerly Twitter) under this account between September 2024 and January 2025.

Despite its small size, it is an authentic dataset with real engagement metrics, making it ideal for small-scale experiments, educational purposes, and exploratory analysis of how post features influence engagement.

📌 Purpose

The dataset is designed to help answer:

Can we predict a thread's engagement rate based on its content, structure, and other posting attributes?

Engagement Rate is defined by X as:

The total number of times a user has interacted with a post. This includes all clicks (hashtags, links, usernames, post expansions), reposts, replies, follows, and likes.

🛠️ Collection Methodology

  • —Data Source: Metrics were collected using X Post Analytics, tracking user engagement, impressions, and other relevant metrics.
  • —Readability Analysis: Grammarly's data was used to compute the Flesch Reading Ease score and other textual analysis metrics.

📊 Features Captured

The dataset includes the following columns:

ColumnDescription
idUnique identifier for each thread
word_countTotal number of words in each thread
reading_time(s)Estimated reading time (in seconds)
readability_scoreFlesch Reading Ease score (higher = easier to read)
posts_per_threadNumber of posts within each thread
topic_complexitySubjective rating of the thread’s topic complexity
media_countNumber of media elements (images, videos, quizzes) per thread
posting_timeTime when the thread was posted (in IST)
post_frequencyNumber of posts made by the account in a week
impressionsNumber of times the thread was viewed
emojisNumber of emojis used within the thread
engagementsTotal user engagements (likes, comments, reposts, follows, etc.)

CSV Header Row: id wordcount readingtime(s) readabilityscore postsperthread topiccomplexity mediacount postingtime post_frequency impressions emojis engagements

🔄 Data Cleaning & Transformation

  • —Basic data cleaning steps were applied.
  • —Consistency checks ensured no missing or corrupted values.
  • —Readability scores were normalized, numeric features standardized where necessary.

📓 Additional Resources

A Jupyter Notebook is available demonstrating:

  • —Exploratory data analysis (EDA)
  • —A simple neural network model built to predict engagement rate.

👉 [Kaggle Notebook Link](https://www.kaggle.com/code/pulkitsahu89/simple-neural-network)

🔍 Potential Use Cases

  • —Investigate the relationship between post characteristics (e.g., content length, readability, media usage) and engagement.
  • —Build machine learning models to predict engagement rate.
  • —Study how readability, timing, and media inclusion affect post performance.
  • —Experiment with small, real-world datasets for educational purposes.

📄 License

  • —License: Apache 2.0
  • —Usage: Publicly available for research and educational purposes.
  • —Commercial Use: Not permitted unless explicitly allowed under the license terms.

📢 Source