Results

***
The explanatory variable log comment count was removed from the model, as its p-value was $0.895$. Our model brings the following linear equation: $$ \log{\hat{y}} = 6.5895 - 0.0442 \cdot published\_in\_weekend + 3.8739 \cdot ratings\_disabled - 0.1624 \cdot comments\_disabled + 0.2835 \cdot \log{likes} + 0.6078 \cdot \log{dislikes} $$ Furthermore, the $R^2-adjusted$ value is $0.824$. As this is fairly close to 1, this means that there is a high degree of correlation between the explanatory variables and the total views for a trending video. The formula states that having all of the explanatory variables equal to zero, the log of views is expected to be $6.5895$. The constants in front of these variables state an increase or decrease of the expected log of the views which are most accurate only when all other variables are kept constant. These values are For the continuous variables log likes and log dislikes it holds that the expected log views for a trending video increases by $0.2835$ and $0.6078$ respectively for every increase of 1 for these variables. If the ratings of a video are disabled the log views is expected increase by another $3.8739$. On the other hand, both the factor of being published in the weekend and the factor of having the comments disabled decrease the expected log views by $0.0442$ and $0.1624$ respectively on their turn. The variables ratings disabled, log likes and log dislikes seem to increase the expected response variable the most as their corresponding coefficients are the largest. A video with many views is expected to be one that has its ratings disabled, has high amounts of both positive and negative ratings as well as not being published in the weekend and not having its comments disabled.