AFL fan community sentiment: Are fans happy or sad?
I generally don’t believe in sentiment analysis. I tend to think it is junk because so much of automated sentiment analysis misses contextual clues, relying too heavily on keywords. Sentiment analysis also tends to give sentiment to bot generated posts and to posts that are neutral. Despite this, I thought it might be a bit interesting to try to do a sentiment analysis of the AFL fan community on Twitter. Are they happy or are they sad? And later, where are the happy fans and where are the sad fans?
The first step in doing that was to collect a whole bunch of AFL related tweets using searchtastic. I’m currently up around 3,400 tweets. (When/If I do the geolocation version, I’ll provide the raw data set.) The second step was to develop a list of AFL specific sentiment related words. In my case, I’m just going with the characteristic of happy and sad to make this easier. My sentiment keyword list is as follows:
| Happy | Sad |
| Best | Worst |
| Win | Spoon |
| happy | lose |
| excited | sucks |
| smile | awful |
| star | sackermanis |
| brownlo | fired |
| medal | suspended |
| victory | fouled |
| club song | sad |
| pride | upset |
| won | fail |
| congrats | heartbreak |
| Purchased | lost |
| lucky | Sold |
| success | Desperate |
| champions | blow it |
| Fit | blew it |
| Riewolt | damn |
| glad | avoid |
| legend | unlucky |
| star | failed |
| brilliant | Akermanis |
| cruised | slammed |
| hope | injured |
| strained | |
| medical | |
| ugly | |
| suspension | |
| crisis | |
| stumble | |
| tribunal | |
| poor | |
| loss | |
| wounded |
The next step was to give a happy or sad label for tweets that included these terms. The last step was to count up how many Tweets were labeled Happy, Happy / Sad, Sad, No Sentiment. I generated the following table:
| Tweets | % | % – None | |
| Happy | 704 | 21% | 55% |
| Happy / Sad | 147 | 4% | 11% |
| Sad | 436 | 13% | 34% |
| No Sentiment | 2128 | 62% | |
| Total | 3415 | ||
| Total | 1287 |
Later, I’m hoping to match this sentiment with the geographic location of the tweets to find out where the pockets of AFL happiness are versus the pockets of AFL sadness are.
