Case Study: YouTube Success Factors — Exploratory Data Analysis

Objective

Identify what video features, like length, tags, publish time, and category, are linked to higher views and engagement.

Data & Approach

• Used YouTube metadata and performance logs from a Kaggle dataset.

• Cleaned data (handled missing values, parsed tags and timestamps, converted categories).

• Ran exploratory data analysis: histograms, scatter plots, boxplots, correlation matrices.

• Compared subgroups: video categories (e.g. education, gaming) and channel sizes (small vs large).

Findings

1. Upload timing: Weekend uploads or after-work hours often yield higher views.

2. Video length: Mid-length videos consistently outperform very short or very long ones.

3. Metadata richness: More descriptive tags and keyword-rich titles correlate with better engagement.

4. Channel consistency: Channels with regular upload schedules build more consistent engagement.

5. Category strategy: Niche or trending categories can perform well if metadata and timing are optimized.

Limitations

• No viewer demographic or watch-time data, so deeper audience insights are missing.

• Analysis shows correlation, not causation.

• External factors like social sharing or promotion aren’t included in the dataset.

Next Steps

• Integrate viewer demographics or watch-time metrics.

• Build machine learning models to predict video success.

• Analyze content features (speech/text, thumbnails, audio/video quality).

• Validate insights via real-world A/B testing of upload strategies.

Conclusion

Video success is strongly influenced by metadata, timing, length, and upload regularity. Creators especially smaller or niche channels, can improve reach and engagement by optimizing these factors.

For full code, charts, and detailed analysis, see the notebook here:👉🏽 YouTube Success Factors: An EDA