Summary: Sentiment analysis is one of the main methods in NLP (Natural Language Processing). Focusing on one of my favorite YouTube channels -- Outdoors Chef Life (https://www.youtube.com/@OutdoorChefLife), this project uses sentiment analysis to analyze the changes in comment polarity for 300+ videos and their comments. Polarity is a measurement of how positive the language is -- 1 means very positive, 0 means neutral, and -1 means negative comments. I then analyzed and visualized the comment polarity by plotting the comment polarity from the oldest video to the latest video, showing the temporal changes in the polarity of Youtube comments across the channel's Youtube history. I also produced a word cloud image based on the video with the highest polarity score (most positive) on its comments.
Data source: YouTube comments extracted from the videos of the channel (https://www.youtube.com/@OutdoorChefLife).
Repository structure:
01_Codes: storing jupyter notebook of python codes. "Youtube_comment_sentiment_analysis_single_video" analyzes one YouTube video using sentiment analysis and produces word-cloud image of the comments from that video. "Youtube_comment_sentiment_analysis_multiple_video" analyzes all 300+ videos by using the loop function, it then produces an interactive Plotly plot to help visualize and track changes in comment polarity across the publication history of the Youtube channel.
02_Results: storing results produced from the python codes, including some CSV files that store the comments of youtube videos, word cloud image, time-series figure, and interactive plotly plot (in HTML format).
Installation Required: Python3 import itertools from youtube_comment_downloader import * import re import pandas as pd import numpy as np import matplotlib.pyplot as plt from IPython.core.interactiveshell import InteractiveShell InteractiveShell.ast_node_interactivity = "all"
#NLP packages from textblob import TextBlob from nltk.corpus import stopwords from wordcloud import WordCloud, STOPWORDS import spacy import nltk
Data visualization:
- Word cloud image
- Time-series visualization of Youtube Comment Polarity
