Playlist Analyser

Background

I have this YouTube playlist which I've been adding every song I like to for almost five years. On the hunt for a dataset I could play with I'd overlooked this; with 2500+ videos (and counting), and it being something I'm directly interested in, it seems the perfect data to get stuck into while I learn more about Data Science.

Now (April 2020) I am doing some exploratory analysis, the first step was creating a data scraper using the Youtube Data API.

My next steps are to improve the quality of the data by doing video-by-video scraping, and to create some pretty visualisations.

Getting Started

Prerequisites

I have developed this using:

• Ubuntu 18.04
• Python 3.6.x

I will try to test this on other operating systems, but don't see any reason it wouldn't work on Windows/MacOS so long as Python 3.x is instaled.

Installation

Installation simple, just clone the repo and run:

pip install reqirements.txt

Runnng the Code

The main runnable script can be run with

python ./scripts/scraper.py

Then the dataframe can be loaded in a python console using

import pandas as pd
df = pandas.read_pickle('/data/video_data.pkl')
df.describe()

Contributing

As its a personal project, and there are probably pre-existing tools that do this better (I haven't checked) I don't expect any contributions. But, if there are any features of interest to anyone, feel free to make a pull request.

I've set the license to GPLv3.