A regression approach for prediction of Youtube views

Lau Tian Rui, Zehan Afizah Afif, R. D. Rohmat Saedudin, Aida Mustapha, Nazim Razali


YouTube has grown to be the number one video streaming platform on Internet and home to millions of content creator around the globe. Predicting the potential amount of YouTube views has proven to be extremely important for helping content creator to understand what type of videos the audience prefers to watch. In this paper, we will be introducing two types of regression models for predicting the total number of views a YouTube video can get based on the statistic that are available to our disposal. The dataset we will be using are released by YouTube to the public. The accuracy of both models are then compared by evaluating the mean absolute error and relative absolute error taken from the result of our experiment. The results showed that Ordinary Least Square method is more capable as compared to the Online Gradient Descent Method in providing a more accurate output because the algorithm allows us to find a gradient that is close as possible to the dependent variables despite having an only above average prediction.


Prediction; Regression; Social media; Social networking; Youtube views

Full Text:


DOI: https://doi.org/10.11591/eei.v8i4.1630


  • There are currently no refbacks.

Bulletin of EEI Stats