Roy Tang

Programmer, engineer, scientist, critic, gamer, dreamer, and kid-at-heart.

Blog Notes Photos Links Archives About

Last week I looked into sentiment analysis, basically it means analyzing text to find out whether it’s “feel good” or “feel bad”. My idea was to use this to analyze my public posts (specifically, on Twitter), to see whether I was trending towards being more positive or more negative. Unfortunately, when I tabulated the data, it was unexciting. No discernable trend of any kind, no plateaus signalling times of stability. Just wild swinging ups and downs as you’d expect from a normal, well-adjusted person. I can’t even get a technical blog post out of it since the process was mind-numbingly simply, I just used the straightforward Python TextBlob library which had a sentiment analysis function.

Sometimes things don’t work out, and that’s fine. Not everything you try is going to be a success. Not all experiments lead to some fantastic discovery. We simply pick ourselves up and move on to the next thing, or the next iteration. If we’re lucky, maybe we learn something new that helps us with the next thing.

It is the nature of human advancement: rarely do things succeed on the first try, in many cases it’s simply a matter of trial and error, of trying a variety of things until something sticks. This is why using an incremental approach is often advisable - trying small things until they succeed and chaining small things into bigger things, so that at any given stage you never fail too much.

Posted by under post at
Also on: twitter / 0 / 242 words