TikTok Video Claim Classification

Overview

This project focuses on building a predictive model to classify TikTok videos as containing either a claim or an opinion. By automating this classification, TikTok can streamline the moderation process, reduce backlog, and prioritize user reports more effectively.

Dataset Information

The dataset consists of:

• Categorical, text, and numerical features.
• Metadata associated with each video.
• A claim_status column indicating whether a video presents a claim or an opinion.

Key Steps

1. Data Inspection & Analysis

   • Reviewed the dataset structure, including numerical and non-numerical variables.
   • Identified missing values in multiple columns, including claim_status.

2. Exploratory Data Analysis (EDA)

   • Analyzed distributions of numerical variables.
   • Identified potential outliers using standard deviation and maximum values.

3. Hypothesis Testing

   • Conducted statistical tests to validate assumptions.
   • Explored correlations between key variables.

4. Regression Modeling

   • Built regression models to analyze relationships between video metadata and claim likelihood.

5. Machine Learning for Video Classification

   • Implemented classification models to predict whether a video contains a claim or an opinion.
   • Evaluated model performance using accuracy, precision, recall, and F1-score.

Results & Insights

• The model successfully differentiates between claims and opinions with high accuracy.
• Insights from EDA help improve content moderation strategies.

Future Work

• Enhance feature engineering for better classification.
• Experiment with deep learning approaches for improved performance.

Author

• Divya - Data Analyst

License

This project is licensed under the MIT License - see the LICENSE file for details.