Insights From The Top 5000 YouTube Channels: Data Cleaning, Analysis, And Visualization
Summary of the Analysis:
This project examines the "Top 5000 YouTube Channels" dataset from Kaggle, leveraging Python's Pandas library for data processing and analysis, and Seaborn for visualization. The workflow begins with data loading and a thorough review to assess the dataset's structure and identify data quality issues. Initial exploration revealed 5000 rows and 6 columns, with only one numerical column ("Video views"). The remaining columns included strings and some missing values, necessitating extensive cleaning to prepare the dataset for deeper analysis.
The data cleaning process addressed missing values, inconsistent formatting, and mixed data types. Missing values, initially marked as "--", were replaced with NaN and subsequently removed. Columns such as "Rank" and "Grade" were reformatted into numerical types for consistency and analysis. Standardized column naming conventions were applied to improve usability. Additionally, a new feature, "Average views," was engineered by dividing "Video views" by "Video uploads," providing an enhanced metric for comparing channel performance.
Exploratory data analysis (EDA) uncovered key insights into the dataset, answering questions about the top-performing channels, grades with the highest video uploads, and grades with the highest subscribers and views. Using groupby operations and correlation matrices, relationships between variables were quantified. For example, the Pearson correlation coefficient revealed the strength and direction of these relationships, offering a deeper understanding of factors contributing to YouTube channel success. These insights, supplemented by visualizations, offer actionable findings for understanding trends among leading YouTube channels.