datatable (@pydatatable) profile picture

Python library for data munging

Mountain View, CA
Joined March 2018
Much scarier than the small number of evil people is the large number who will fall in line with whatever's fashionable. When kindness is in fashion, they're kind. When a form of bigotry becomes cool, they jump on board. When it's popular to condone violence, they're all for it.
189
1,502
175
9,328
datatable retweeted
Replying to @waitbutwhy
I never realized this consciously before now, but you have to be independent-minded to be good. Otherwise you'll be drawn into doing bad things out of conformism.
94
292
30
2,673
Some people have asked why I am tweeting about “politics”. The reason is that I have vulnerable family members and they and others like them may not survive this epidemic in large part due to the Trump administration’s gross incompetence, cruelty, and/or criminality.
16
27
3
348
I really appreciate some of the nice little aesthetic touches in #pydatatable. Still not as mature (or familiar) as #rdatatable, but I'm enjoying it so far. Thanks, @h2oai!
1
1
9
datatable retweeted
Our own @pstetsenko & Oleksiy Kononenko presenting Data.Table #H2OWorld. Watch the live stream from Godel-Pauling Stage here: bit.ly/H2OSFLive
2
4
This is a super cool resource: Papers With Code now includes 950+ ML tasks, 500+ evaluation tables (including SOTA results) and 8500+ papers with code. Probably the largest collection of NLP tasks I've seen including 140+ tasks and 100 datasets. paperswithcode.com/sota
38
1,087
74
2,480
If an eye doctor looked at a retinal photo, the chance of getting gender correct would be 50-50. But deep learning training led to an AUC of 0.97 @pearsekeane pointed out how striking this is @JeffDean at a recent #AI @GoogleAI meeting; data: nature.com/articles/s41551-0… @NatBME #OA
60
655
203
1,271
Apparently, Slack doesn't have any plans to implement code highlighting for back ticks. Please retweet if you want code highlighting with ```js or ```php or ```py or anything else that GitHub supports. Let's show Slack how badly we need this.
Replying to @tarasm
Got it. We don't have immediate plans to support specific languages when using backticks, but we'll pass the suggestion along!
22
321
10
397
datatable retweeted
My talk last week on data.table for R and Python, and updated-daily benchmarks. #rdatatable @pydatatable youtu.be/Ddr8N9STSuI
1
28
1
56
Our own benchmarks show that the sentinel and bitmask methods go neck and neck with each other (except when the number of NAs is very low, in which case the bitmask method has an upper hand). Check the results at github.com/st-pasha/microben…
New post! "Is it time to stop using sentinel values for null (NA) values?". The great NaN vs. bit/byte-mask debate wesmckinney.com/blog/bitmaps… @ApacheArrow #pydata #rstats
2
Very excited that Facebook has just become a Principal Sponsor of the Python Software Foundation: python.org/psf/sponsorship/s…
4
43
4
193
datatable is at 1000 commits today! 🎉 On an unrelated note, "Round Number Bias" is a term in psychology for humans' proclivity to pay special attention to numbers that are round.
1
7
"You can get a good idea on the corner for $4 with your latte. What matters is how well you execute." – Steve Jobs, yelling at us after the MobileMe debacle I wish more people would talk about this. Last I checked, this quote got lost, so I'm telling you now, since I was there.
7
187
16
823
Folks, the #rdatatable spirit is now available in Python too! It's in Alpha stage, so you are encouraged to report bugs and submit proposal! Special thanks to @MattDowle for his effort in developing intuitive and efficient libraries for #DataScience Link-> github.com/h2oai/datatable
3
4
Selecting columns of a certain type from a Frame is easy, the type itself can be used as the selector: ``` DT[..., str] ```
1
5
#rstats #rdatatable dev version now has a simple typo checker to check for potential misspellings in your i query. Implementation is quite rudimentary for now -- all feedback is welcome/appreciated! Give it a spin! github.com/Rdatatable/data.t…
1
5
10