Built forecasting models to predict video view time based on customers’ activities by utilizing Amazon Prime dataset
Findings:
- From the figure above, the most important feature is the number of imdb_votes likes, that is, the more people like it, the more popular the movie will be;
- The second important feature is category_position and the third important feature is horizontal_position. That is the "Golden Triangle" because if a movie appearing in the upper left corner, then its click-through rate will become higher;
- At the same time, the least important feature is n_years (the time when the movie appeared), that is, the year the movie came out is not very important.
