Motivation:¶

Problem¶

Is there a 'formula' that allows videos to be posted on trending page? Is there a certain way the video should be named? Does it have to be certain length or it has something to do with correlation of views and likes over a certain time period?

Solution¶

YouTube is the biggest global online video sharing and social media platform. The scope of their platform allows for a rich dataset of videos that went "trending" and see what were the characteristics of these videos in terms of The goal of this project is to identify and use a relationship between YouTube's video features (likes, dislikes, views, tags, title and etc)

Impact¶

If successful, the classifier might be used by creators to put out videos that have high likelihood of being put on trending page, increasing their popularity, therefore incresing their income. It can also be used by companies that want to advertise their product through creator by requesting optimal format (that will be discovered) that will allow their ad to be more effective (to be seen by more people).

One (very) negative outcome of such a classifier is that it may encourage creators to follow rules instead of allowing themselves to be creative and put out less authentic content which their subscribers might not enjoy.

Dataset¶

Detail¶

We will use a Kaggle Dataset to observe the following features for each video:

video_id - id of the video
trending date - when video got on trending page
title - title of the video
channel_title - title of publisher's channel
category_id - category of that video
publish_time - the time when it was published
tags - tags that were put under that video
views - amount of views
likes - amount of likes
dislikes - amount of dislikes

Potential Problems¶

This dataset does not have video length that I think plays part in whether video will be put on Trending page. We will be using US vides data only since I want to do analysis of titles and it will be hard to do that if titles will be on different languages, that might skew our analisis results. But if needed, there are data sets for other countries as well.

Method:¶

We will look for correlation between views and reactions (likes and dislikes), frequency of categories that get on trending page the most. Find the best time for a video to be uploaded that will increase likelyhood of it being on trending page. The same thing will be done with tags. We also will analise titles and see if there is a correlation of certain words in the title and trending page.

What it takes for a YouTube video to become Trending.¶