Data VisualizationAugust 26, 20232 min read

YouTube Statistics Analysis

Problem Statement

What separates the world's top YouTube channels from the rest? Using a Kaggle dataset of 995 top global YouTube channels, this analysis identifies the features, categories, and geographic patterns most strongly associated with subscriber growth and estimated earnings.

Methodology

Data Preprocessing

The dataset from Kaggle was relatively clean, requiring minimal handling of missing values and outliers. Basic statistics were computed including averages, standard deviations, and correlation matrices.

Correlation matrix heatmap showing relationships between YouTube channel features — Correlation matrix revealing several highly correlated features in the dataset.

Results

Feature Importance

A Random Forest Classifier was used to model the features associated with YouTube success (defined as a Top 50 rank).

Content Categories

Music channels on YouTube generate disproportionately more views per upload compared to other categories. This makes sense as consumers rewatch music videos far more than time-sensitive content like News & Politics.

Uploads vs. video views per category showing music dominance — Uploads vs. video views per category - music channels have an outsized views-to-uploads ratio.

Regional Influencers

Subscribers & Earnings

There is a statistically significant association between subscriber count and monthly earnings, suggesting aspiring creators should prioritize subscriber growth.

Scatter plot showing correlation between subscribers and monthly earnings — Subscriber count strongly correlates with monthly earnings.

Next Steps

Examine historical YouTube statistics to identify trends in consumer taste over time
Compare YouTube statistics with other platforms to understand broader digital media success factors
Apply more sophisticated class imbalance techniques (SMOTE, ensemble methods) for improved modeling