I played with doing something similar, where my features were based on Edgic charts I found online[1]. This ends up implicitly factoring in people’s _perceptions_ of the players/editing, which can be useful. Ultimately it was just a bit of a toy still, somewhat better than random.
Like other people mention, Survivor is edited with a lot of intention, I think if you really want to predict winners you need to somehow capture cues from the editors.
I wonder if you could derive per-player 1st and 2nd person sentiment from transcripts of the episodes. Could be interesting to factor that in.
[1]: Example - <a href="https://www.reddit.com/r/Edgic/s/2v0nA93odi" rel="nofollow">https://www.reddit.com/r/Edgic/s/2v0nA93odi
guiand · · focus · HN ↗
I played with doing something similar, where my features were based on Edgic charts I found online[1]. This ends up implicitly factoring in people’s _perceptions_ of the players/editing, which can be useful. Ultimately it was just a bit of a toy still, somewhat better than random.
Like other people mention, Survivor is edited with a lot of intention, I think if you really want to predict winners you need to somehow capture cues from the editors.
I wonder if you could derive per-player 1st and 2nd person sentiment from transcripts of the episodes. Could be interesting to factor that in.
[1]: Example - <a href="https://www.reddit.com/r/Edgic/s/2v0nA93odi" rel="nofollow">https://www.reddit.com/r/Edgic/s/2v0nA93odi