Engagement Metrics Analysis of a Perfume Youtube Channel Using YouTube API

Overview

What better way to discover new products than through a recommendation? It's pre-vetted and, if you're lucky, it includes a detailed breakdown of the product's pros and cons. Nowadays, if we're considering buying a new sunscreen, lotion, or perfume, it's likely because we've seen it highlighted by an influencer on social media. For this project, I analyze Monika Cioch's YouTube channel, which is largely dedicated to reviewing, discovering, and recommending perfumes. This analysis aims to understand her viewers' preferences by examining engagement metrics such as views, likes, and comments count to determine which of her videos performs the best.

Data Context

For the analysis, each video was categorized into a specific group

Category Name	Description
recommendation	in-depth description of perfume(s)
review	first impression of a perfume or perfume comparison
informative	in-depth breakdown of a specific perfume or videos on how to choose a perfume
fun	non-perfume related videos for entertainment
fashion	non-perfume related videos about fashion
shorts	Youtube shorts

Analytics

Key Insights

• Monika's audience prefers review videos. Although they make up just 11.7% of her content, review videos receive the highest number of views and comments, with likes trailing only slightly behind Youtube shorts. To note, it is also the category with the most outliers based on the violin plot.
• Based on all three engagement metrics, longform content tend to generally outperform shortform content
• The top 4 videos that consistently rank in the top 10 are about the types of perfumes that men would find appealing on women 👀

Future Work

• Comments analysis
• Classify videos using more specific categories
• Audience demographics analysis
• Include the variable of time: Are the best performing videos the ones that have been up the longest?

Repository Contents

Folder: data

All data used to complete project

File Name	Data Description
video_info.csv	raw data from Youtube API
video_info_with_categories.csv	added new column named 'category' to classify videos
video_info_cleaned.csv	final cleaned data

Python File: youtube_api_data_collection.py

• Extracts engagement metrics from a perfume YouTube channel
• Utilizes the YouTube API for data retrieval
• Saves the data in csv format

Jupyter Notebook: data_transformation_fragrance_youtube.ipynb

• Jupyter notebook with annotations detailing each stage of preprocessing data from Youtube's API
• Data exploration process
• Plotly visualizations

Folder: assets

All assets used to complete project

• pictures
• graphs
• tables