Tweets

	Randy Olson @randal_olson	12h
	How I hope the rest of the story goes: #Trump & co are proven to have conspired w/ Russians to meddle in the US election & are sent to jail. Joe Biden elected President in a special election. Blue Midterm happens. Actual #MAGA. There is hope. #Mueller twitter.com/randal_olson/s…
	View details ·

	Randy Olson @randal_olson	12h
	It seems the story so far is: Russians get aggressive with their neighbors. US imposes economic sanctions, conspires to gut oil prices. Russian economy starts declining. #Russia responds by investing in cyber and political warfare (propaganda) w/ the US. (from Reddit) #Mueller
	View details ·

	Randy Olson @randal_olson	17h
	Replying to @Michielstock @Ronald_vanLoon and 5 others Yes, and always compare to a baseline model that predicts only the majority class.
	View conversation ·

	Randy Olson @randal_olson	18h
	Replying to @Ronald_vanLoon @TDataScience and 4 others #3 is more nuanced than that. What if one of the classes is 80% of the data? Then 80% accuracy is a failure.
	View conversation ·

	Randy Olson @randal_olson	Feb 15
	Replying to @noooooomad @SciPyConf The past, present, and future of Automated Machine Learning.
	View conversation ·

	Randy Olson @randal_olson	Feb 15
	Replying to @Skilenstein Sometimes there’s still room if you can do something better than the existing solutions.
	View conversation ·

	Randy Olson @randal_olson	Feb 15
	Replying to @kcunning I knew about the existing solutions but thought I could do better. I was wrong this time. 😅
	View conversation ·

	Randy Olson @randal_olson	Feb 15
	This morning has been a lesson in, "Stop trying to roll your own smart solutions to a common #programming problem when other smart people have already solved that problem and released those solutions as a #Python package" for me.
	View details ·

	Randy Olson @randal_olson	Feb 15
	Useful guide for interpreting software engineering job postings. #programming reddit.com/r/ProgrammerHu… pic.twitter.com/Esao8pNS8H
	View photo ·

	Randy Olson @randal_olson	Feb 15
	Reminder that using eval in your #Python code can be very dangerous IF you're using it to run code with any information from untrusted sources. #programming #netsec reddit.com/r/Python/comme… pic.twitter.com/EFawBq2DS2
	View photo ·

	Randy Olson @randal_olson	Feb 15
	#Execution rates in #G20 members (2016) #dataviz The number of executions in #Russia must be a better-kept state secret. reddit.com/r/dataisbeauti… pic.twitter.com/foQLPQmJqP
	View photo ·

	Randy Olson @randal_olson	Feb 15
	Replying to @ash_chakraborty Probably. Amphetamine is the new replacement app for Caffeine on macOS, btw. Offers almost all the functionality of this command line tool except being able to specify a PID.
	View conversation ·

	Randy Olson @randal_olson	Feb 14
	Replying to @Stevezieglerva Running local code overnight.
	View conversation ·

	Randy Olson @randal_olson	Feb 14
	I can't believe I'm just learning about the caffeinate command today. The -w parameter makes it more useful than any similar tool I've seen before. pic.twitter.com/tVskYLEhCK
	View photo ·

	Randy Olson retweeted
	Andrej Karpathy @karpathy	Feb 14
	"Deep Reinforcement Learning Doesn't Work Yet" alexirpan.com/2018/02/14/rl-… great read, hits a lot of points I've also come to realize over last ~2 years. 70% is a vast understatement.
	View summary ·

	Randy Olson @randal_olson	Feb 14
	Replying to @SharpestMindsAI @jeremyphoward and 2 others Makes sense!
	View conversation ·

	Randy Olson @randal_olson	Feb 14
	Replying to @rasbt @SciPyConf Already booked up for July?
	View conversation ·

	Randy Olson @randal_olson	Feb 14
	Got my @SciPyConf talk proposal submitted just in time - phew. Fingers crossed!
	View details ·

	Randy Olson retweeted
	Randy Olson @randal_olson	Feb 11
	An animated map of the #migration routes of 118 species of #birds in North America. #dataviz audubon.org/news/see-milli… pic.twitter.com/QublGt4sqw
	View photo ·

	Randy Olson retweeted
	Leland McInnes @leland_mcinnes	Feb 12
	Our paper on UMAP, a faster alternative to t-SNE, is now up on arXiv! The paper provides a more detailed account of the theoretical underpinnings of the algorithm, as well as performance benchmarks. arxiv.org/abs/1802.03426 pic.twitter.com/kzUng6ZtNC
	View photo ·