willow_hiller
Well-Known Member
Someone on Reddit conducted a comment sentiment analysis for comments that contained "Elon Musk."
Alright, before anyone draws any more conclusions on the graph I posted, I was a bit suspicious because that OP didn't post any code or methods except for the "VADER" sentiment analysis, so I attempted to create the same chart for the search term "tesla" since the beginning of the year. Long story short, that chart must have a HUGE amount of smoothing applied to it.
My method: Take the 100 most upvoted comments per day that contain the word "tesla" from 2020-01-01 until today. Calculate the nltk VADER sentiment compound score for each comment, and take the average for the day. Below is the scatter plot of this method, where the compound score can range from -1 to 1:
So overall positive, except for 3 days this year (including May 1). Is there a trend there? I don't think so, but you can force a trend on it if you calculate the LOESS: