John Healy

[John Healy @JohnDataHealy]

@pedramamini Hi Pedram, it looks like the monacle github page isn't available (404). Did it get removed or set to private?

[Cathrine Sant @cathrine_sant]

When you tell an 'omics lab to dress up for Halloween.. @doctorcorces

[John Healy @JohnDataHealy]

@cathrinepet @doctorcorces @ZachGardell @irischin9 I never thought that our UMAP algorithm would ever inspire a Halloween costume. Much less an awesome one. Well done!

[Chris Huston @chuston_ai]

Holy crap - I am going to use HDBSCAN: https://t.co/zvxUDQ0DvM. I wish I’d have found this in 2019 where I was lost wandering in a big data forest of clustered matrices and struggled to ferret out what was real. Now, do I reopen that crypt? As if I could sleep if I didn’t.

[Ramon Massoni Badosa @rmassonix]

After 9 months of PhD I have finally found a palette for a 6-color tSNE/UMAP that is divergent, colorblind-friendly and readable in black&white. God bless the queen!

queen <- c("#CD2C14", "#F4B4A3", "#EEE2B9", "#4FAA52", "#1A40AD", "#8C2EA9")

[John Healy @JohnDataHealy]

@ToferC @DaveConabree Have a look at Brandon Rhodes old intro to pandas course that he gave a pydata. It is probably more detailed than you want but it has very well designed assignments and wonderfully smooth learning curve. We've used it as a basis for an intro course a number of times now.

[Dave Conabree @DaveConabree]

I am tempted to order about a thousand of these t-shirts....

[Dave Conabree @DaveConabree]

@worldofabe Get out of your office and get to know people on a personal level, every day. It's the connections formed in those genuine relationships that support the conversations on the tougher stuff.

[Amjad @AmjadTaleb]

Using @BokehPlots for data visualization is such a joyful experience.

I made an application to help colleague scientists have fun while easily and quickly exploring their data.

https://t.co/5VZ0l6aZbQ

[Alexandra Petri @petridishes]

why is the correct approach to reading tagged as “neutral evil”

[Alexandra Petri @petridishes]

from Jabba the Hutt's perspective Star Wars is the story of a guy who owed him money and then instead of paying had his friends murder him

[Alexandra Petri @petridishes]

moby dick would also like to state that what ahab saw as a mutual nemesis relationship was just ahab stalking him with a boat for years

[Rongxin Fang, Ph.D. @RongxinFang]

SnapATAC now includes multiple dimensionality reduction techniques JDA, LDA (cis-topic) and LSA, LSA-logTF, allowing for finding consensus clusters between methods! Inspired by @ahill_tweets recent post. #rstats #singleATAC #MachineLearning

Notebook: https://t.co/vEv6UMWhWI

[Ryan Abernathey @rabernat]

@JuliusBusecke @ProjectJupyter @globus I would encourage you to escalate this issue and see if you can get a change in policy. Soon will be the Jupyter for HPC workshop (https://t.co/KghJrrIPlS). and there is growing awareness of the value Jupyter brings to HPC platforms. Convince admins they are behind the times!

[Steve Astels @sastels]

It’s taken me many years to realize that if people are using my product “wrong”, the problem is with the product, not the people.

[James Bednar @JamesABednar]

Do you need to plot a large number of categories at once? Typical categorical colormaps have at most 10 or 20 colors, but Colorcet now provides several collections of 256 perceptually distinct colors, so that you never have to repeat colors again! https://t.co/PSrbhtYC0j

[Adam J Taylor @adamjtaylor]

Safe to say that the inverse transform from the 0.4dev branch of @leland_mcinnes's already powerful #UMAP is witchcraft. Without any optimisation it recovered images from from fashion MNIST with mean cosine similarity of 0.93.

[micah.fyi @micahstubbs]

a nice blog post from @ceshine_en exploring the new NVIDIA #RAPIDS GPU library. shows a 15x Speedup on #UMAP vs CPU!

📄 https://t.co/HPtOpx3VeJ

[Francesca Guiducci @engfragui]

Speaker addresses the audience: “What debugger do you use when you write Python code?”
From the audience: “Print statements?”
#PyCon2019

[Simon Anders @s_anders_m]

Sleepwalk, our R package for interactively exploring and comparing t-SNE, UMAP or other dimension-reduced embeddings, is now on CRAN, and a preprint about it is on BioRxiv. Preprint: https://t.co/zy2AF8iHVD ; web page: https://t.co/7znm7mLdEB