By the way, I see that you have fitted an quadratic polynomial to the “number of chapters as a function of time” dependency. Can you eloborate on the reasons you used quadratic polynomial as opposed to sqrt(x) or log(x), the predicted date of the next update and the degree of trust such model deserves?
EDIT: It appears that I had a brainfart and mixed up plots. Nevermind then.
PS: Also, I suspect that taking total length of reviews into account can reveal something new about distinction between “hard fans” and “populists”.
PS: Also, I suspect that taking total length of reviews into account can reveal something new about distinction between “hard fans” and “populists”.
Possibly, and maybe also try employing some ‘sentiment analysis’ to estimate review favorability, yeah.
But I probably won’t bother with this because I’ve answered the questions I was interested in, and doing those extra analyses would be much more work—not only do I not know how to use any of those tools, I wouldn’t even be able to reuse any of my existing code, eg. the download code for downloading entire reviews would be more complicated (I would have to do real HTML parsing instead of just grabbing single lines with user/chapter/date in a stereotypical format).
That was very interesting, thank you.
By the way, I see that you have fitted an quadratic polynomial to the “number of chapters as a function of time” dependency. Can you eloborate on the reasons you used quadratic polynomial as opposed to sqrt(x) or log(x), the predicted date of the next update and the degree of trust such model deserves?
EDIT: It appears that I had a brainfart and mixed up plots. Nevermind then.
PS: Also, I suspect that taking total length of reviews into account can reveal something new about distinction between “hard fans” and “populists”.
Possibly, and maybe also try employing some ‘sentiment analysis’ to estimate review favorability, yeah.
But I probably won’t bother with this because I’ve answered the questions I was interested in, and doing those extra analyses would be much more work—not only do I not know how to use any of those tools, I wouldn’t even be able to reuse any of my existing code, eg. the download code for downloading entire reviews would be more complicated (I would have to do real HTML parsing instead of just grabbing single lines with user/chapter/date in a stereotypical format).