class: center, middle, inverse, title-slide # Exploring Factors of Popularity in U.S. Youtube Videos ## What makes a Youtube video popular? ### CharlieBitMyFinger
Jesse Solinger, Oliver Wan, and Riley Burns ### Bates College ### 2022-04-20 --- class: left, top ## Research Question What makes a Youtube video popular? And what factors contribute to videos' popularity? Youtube maintains a list of top trending videos posted on their platform each year. According to *Variety* magazine, Youtube does not base their top trending video list solely on views. They measure user interactions (number of views, shares, comments and likes), which all impact video popularity. Keeping this in mind, we will explore what specific factors contribute to popularity and how we can predict a video's eventual popularity. --- class: left, top # Trending Youtube Video Statistics We aim to examine what makes a Youtube video popular using a dataset from Kaggle.com that contains a daily record of the top trending Youtube videos and their statistics. The dataset includes the video title, channel title, publish time, tags, views, likes and dislikes, description, and comment count. The data also contains data from 10 different countries, but for the purpose of this project, we will stick to data from the U.S. It's also important to note that this dataset uses Youtube data from 2017 to 2018. <div class="figure" style="text-align: center"> <img src="img/wired-youtube.jpg.webp" alt="world map of youtube logo on top to demonstrate the topic of our presentation." width="35%" /> <p class="caption">Image Credit: url(https://www.wired.co.uk/article/youtube-migration-videos)</p> </div> --- # How Do Likes Impact Views? .pull-left[ <img src="presentation_files/figure-html/likes-views-1.png" title="Scatter plot with linear model of relationship looking at Views vs. Likes. This graph demonstrates how increased likes are correlated to higher viewer count." alt="Scatter plot with linear model of relationship looking at Views vs. Likes. This graph demonstrates how increased likes are correlated to higher viewer count." width="100%" /> ] .pull-right[ <img src="presentation_files/figure-html/likes-views-faceted-1.png" title="Faceted scatter plots of Likes vs. Views to break down original graph by category to isolate each genre of video posted. These graphs demonstrate relationship between Likes vs. Views by category." alt="Faceted scatter plots of Likes vs. Views to break down original graph by category to isolate each genre of video posted. These graphs demonstrate relationship between Likes vs. Views by category." width="100%" /> ] --- # How Do Dislikes Impact Views? .pull-left[ <img src="presentation_files/figure-html/dislikes-views-1.png" title="Scatter plot with linear model of relationship looking at Views vs. Dislikes. This graph demonstrates how increased dislikes are correlated to higher viewer count." alt="Scatter plot with linear model of relationship looking at Views vs. Dislikes. This graph demonstrates how increased dislikes are correlated to higher viewer count." width="100%" /> ] .pull-right[ <img src="presentation_files/figure-html/dislikes-views-faceted-1.png" title="Faceted scatter plots of Dislikes vs. Views to break down original graph by category to isolate each genre of video posted. These graphs demonstrate relationship between Dislikes vs. Views by category." alt="Faceted scatter plots of Dislikes vs. Views to break down original graph by category to isolate each genre of video posted. These graphs demonstrate relationship between Dislikes vs. Views by category." width="100%" /> ] --- # How Does Time Impact Views? (Date Published) .pull-left[ <img src="presentation_files/figure-html/views vs date published-1.png" title="Scatter plot displaying the relationship between youtube views and date posted. Videos posted from November through till June tend to receive higher viewership." alt="Scatter plot displaying the relationship between youtube views and date posted. Videos posted from November through till June tend to receive higher viewership." width="100%" /> ] .pull-right[ <img src="presentation_files/figure-html/views vs date published facet-1.png" title="Faceted scatter plot of previous graph broken down by category. Graph aims to breakdown and provide clearer insight on specific genres of videos and the relationship between views and date posted." alt="Faceted scatter plot of previous graph broken down by category. Graph aims to breakdown and provide clearer insight on specific genres of videos and the relationship between views and date posted." width="100%" /> ] --- #How Does Time Impact Views? (Months and Minutes) .pull-left[ <img src="presentation_files/figure-html/views vs month-1.png" title="Polar graph displaying the relationship bewteen views and upload month. This graph aims to visually display views in relationship to month on a circular scale." alt="Polar graph displaying the relationship bewteen views and upload month. This graph aims to visually display views in relationship to month on a circular scale." width="100%" /> ] .pull-right[ <img src="presentation_files/figure-html/views vs hour-1.png" title="Polar graph displaying the relationship bewteen views and upload hour. This graph aims to visually display views in relationship to hour on a circular scale." alt="Polar graph displaying the relationship bewteen views and upload hour. This graph aims to visually display views in relationship to hour on a circular scale." width="100%" /> ] --- # How Many Videos are Produced Within Each Category? <img src="presentation_files/figure-html/category-videos-1.png" title="Barplot displaying the number of videos in each video category. Aims to demonstrate which category has the most number of videos." alt="Barplot displaying the number of videos in each video category. Aims to demonstrate which category has the most number of videos." width="70%" style="display: block; margin: auto;" /> --- # What is the Relationship Between Views and Category? <img src="presentation_files/figure-html/category-v-views-1.png" title="Barplot displaying the relationship bewteen YouTube Category and Views. Barplot aims to show which categories have the highest viewcount." alt="Barplot displaying the relationship bewteen YouTube Category and Views. Barplot aims to show which categories have the highest viewcount." width="70%" style="display: block; margin: auto;" /> --- # Conclusions - The slope of dislikes vs views varies significantly from category to category - The slope of views vs likes is not as variable from category to category - Date published does not seem to correlate with video popularity at all - However, some videos in April and May published tend to garner a ton of views - A significant portion of Youtube videos are categorized as music or entertainment - Music and entertainment videos are also some of the most popular videos on Youtube - Videos in channels that have a ton of videos tend to have more views - Videos in categories that have a ton of videos tend to have more views --- class: left, top # Limitations - Potential data manipulation - Mitchell J (Kaggle username) is the source for our data… what if Mitchell J messed with the dataset before posting it on Kaggle - Mitchell J scraped the data from Youtube… what if Youtube altered its own data before making it publicly available? - Youtube guidelines for collecting data - What counts as a view? - The data is only representative of viewers in the US - people use Youtube in many other countries as well --- class: left, top # Future Research - Examine up-to-date Youtube data - Examine misinformation related to democracy / government censorship - Examine data from other platforms (TikTok, Instagram, Facebook, etc.) - Examine Youtube advertisement data / Youtube views by geography - A thorough analysis of many years of Youtube data could dive into how changing Youtube monetization rules have affected the ways that channels create content --- class: left, top # Questions? <div class="figure" style="text-align: center"> <img src="img/210524-charlie-bit-my-finger-se-1035a.webp" alt="Image Credit: url(https://www.nbcnews.com/pop-culture/viral/charlie-bit-my-finger-youtube-video-sells-nft-760-999-n1268322)" width="65%" /> <p class="caption">Image Credit: url(https://www.nbcnews.com/pop-culture/viral/charlie-bit-my-finger-youtube-video-sells-nft-760-999-n1268322)</p> </div> --- #Extra Credit Regression ``` ## parsnip model object ## ## Fit time: 570ms ## ## Call: ## stats::lm(formula = views ~ category + publish_time + likes + ## comment_count + dislikes, data = data) ## ## Coefficients: ## (Intercept) categoryComedy ## -1.138e+07 -1.456e+06 ## categoryEducation categoryEntertainment ## -1.233e+06 -6.326e+05 ## categoryFilm & Animation categoryGaming ## -3.883e+04 -7.383e+05 ## categoryHowto & Style categoryMusic ## -1.146e+06 -1.587e+06 ## categoryNews & Politics categoryNonprofits & Activism ## -7.034e+05 -4.014e+06 ## categoryOther categoryPeople & Blogs ## -8.328e+05 -1.211e+06 ## categoryPets & Animals categoryScience & Technology ## -8.405e+05 -5.948e+05 ## categorySports categoryTravel & Events ## -4.395e+05 -5.744e+05 ## publish_time likes ## 8.254e-03 3.608e+01 ## comment_count dislikes ## -9.953e+01 8.357e+01 ``` --- #Extra Credit Revision <img src="presentation_files/figure-html/views-v-likes-revised-1.png" title="Scatter plot with linear model of relationship looking at Views vs. Likes. This graph demonstrates how increased likes are correlated to higher viewer count." alt="Scatter plot with linear model of relationship looking at Views vs. Likes. This graph demonstrates how increased likes are correlated to higher viewer count." width="80%" /><img src="presentation_files/figure-html/views-v-likes-revised-2.png" title="Scatter plot with linear model of relationship looking at Views vs. Likes. This graph demonstrates how increased likes are correlated to higher viewer count." alt="Scatter plot with linear model of relationship looking at Views vs. Likes. This graph demonstrates how increased likes are correlated to higher viewer count." width="80%" /> ```