Re:What we learned from designing 200 pitch decks
by Animats on Thursday August 13, 2015 @03:41PM
(#10056693)
Somebody should run a recursive deep ANN on pitch decks to optimize for money raised, and see what it comes up with as the optimal pitch deck.
[ Parent
| on HN ]
Re:What we learned from designing 200 pitch decks
by pcrh on Thursday August 13, 2015 @03:53PM
(#10056776)
See this thread:
Re:What we learned from designing 200 pitch decks
by shoo on Thursday August 13, 2015 @11:52PM
(#10058537)
Arguably one might need to capture a bit more data describing the context and outcome associated with each recorded "deployment" of each deck.
With only a couple of hundred samples, and a highly complex process with many poorly understood interacting factors, and loads of missing information, throwing neural networks at it sounds like a fun exercise in overfitting!
Could probably get further by getting hold of a few experts who have strong prior beliefs about what factors may be relevant, and then carefully running some statistical analysis / additional experiments!
"statistics" and domain expertise - not trendily branded, probably still a great idea when you merely have a few hundred samples in a crazily ill-defined high dimensional space.