1
@Quesada Yes to all. 1) tooling landscape around ML is starting to exist, but still somewhat experimental and fragmented.
Back to archive
Thread
3 tweets
2
@dataScienceRet @Quesada 2) pretrained models and data exists for some problems, but if you want to do something new, you need to „bootstrap data“, which is work and a bit of „dark science.“
3
@dataScienceRet @Quesada 3) I wouldn‘t say they are afraid but you still need to know what you are doing and non-DS people lack the statistics intuition to make sure it works out of sample.