4 ms·
>> Interpretability obviously doesn't hurt accuracy. Why do you say this? From what i have seen it certainly can and does. For some industries finding a trade
by 1_over_n 7y ago
>> Interpretability obviously doesn't hurt accuracy.
Why do you say this? From what i have seen it certainly can and does. For some industries finding a trade off where the magic is.
https://link.springer.com/article/10.1007/s10664-018-9638-1 https://link.springer.com/article/10.1007/s10664-018-9638-1
https://www.oreilly.com/ideas/predictive-modeling-striking-a-balance-between-accuracy-and-interpretability https://www.oreilly.com/ideas/predictive-modeling-striking-a...
- dchichkov 7y agoYou can take nearly any black box model and add some interpretability into it, without modifying this model. No modifications to the model, so accuracy stays the same.