Youtube-Views-Prediction

Supervised Machine Learning Model: YouTube Views Prediction

Summary

1. Objectives: Predicting YouTube views using ML models

2. From modeling results that have been carried out on the YouTube Statistics dataset to predict the number of views, we can conclude that the best model is Random Forest which gives an R2 Score: of 0.95 and small MAE and RMSE values (0.39 and 1.2). This means that the model shows good performance in predicting targets.

3. The second best model is the Decision Tree model with R2 Score: of 0.95 and MAE and RMSE values which are not much different from Random Forest model (0.4 and 1.61). This shows that a suitable method is using ensemble learning-Bagging.

4. Recommendations that can be given to increase the number of views are:

   • From publication time, we can publish the video on weekends, which has an influence on increasing the number of views. If this is done, we will most likely obtain significant results :
     A. The number of views will increase
     B. If the number of views increases, the number of likes, dislikes, and comments will automatically increase.
     C. If there are more and more of these things, the YouTube algorithm will detect the video as a trending video.

   • Don't disable comments and ratings. From the data analysis, it can be seen that if comments or ratings are disabled, then someone's interest in watching the video will also decrease, so the number of interactions with the video will also decrease.

   • Category of content also affects the number of views. Choose the category that the audience often likes, then adjust the tags (both title tags and number of tags) with good proportions.