@_fernando_rosas@rlmcelreath @philipcball here is a great talk by J Crutchfield from a couple years ago, where he talks about e-machines, emergence and demonology https://t.co/H2ChQOaVyt (better quality link in comments) may be worth checking out if anyone is interested to learn more about the topic
great article!
@JoachimSchork any idea how to detect Simpson's paradox in large p small n scenario?
I find that it is reasonable to manually check for it when there are tens of variables, but it becomes nigh impossible when there are 100s-1000s
@davisidarta@mihirraov@PNASNews i feel like the omics field in general needs more timepoints and higher quality datasets, instead of even more complex preprocessing pipelines
@davisidarta@PNASNews bold!
especially re "if this was just an honest mistake. It isn't. The community was deceived by a few".
What makes you say that? I thought a lot of the preprocessing methods that became de facto norm were developed to address practical limitations at the time, eg p>>n data