Anna Neufeld

52 posts

Anna Neufeld

@AnnaCNeufeld

Statistics PhD Student @UW. Enthusiastic about mountains and p-values.

Katılım Ekim 2020

122 Takip Edilen270 Takipçiler

Anna Neufeld@AnnaCNeufeld·26 Tem

@MHemingNeuro @ScienceScottT @tangming2005 @hoffman_steven Also, we have more information about using count splitting more picking the number of clusters in our new preprint, which expands the original count splitting paper to the negative binomial setting: arxiv.org/abs/2307.12985

English

1.2K

Anna Neufeld@AnnaCNeufeld·26 Tem

@MHemingNeuro @ScienceScottT @tangming2005 @hoffman_steven Yes! We have one in the data thin package! I also just added one to the count split tutorials package on simulated data! I am working on updating the "real data" Seurat/Monocle/etc. tutorials to include more on model evaluation! anna-neufeld.github.io/countsplit.tut…

English

Ming "Tommy" Tang@tangming2005·24 Tem

chatomics! Is my single cell cluster significant? New blog post: scRNAseq clustering significance test: an unsolvable problem? buff.ly/472nMlz My first try of nature.com/articles/s4159… #rstats #singlecell

English

146

21.6K

Anna Neufeld@AnnaCNeufeld·8 Haz

@Spottnik Not for brunch, but I recommend Leny’s in tangletown!! For next Thursday evening (June 15!)

English

204

Greg Spotts@Spottnik·8 Haz

I’m looking for ideas for time & location for a #Summerof62 brunch meetup this Sunday June 11 Talk to me

English

Mike Wu@wu_biostat·6 Nis

Best part of being a @UWBiostat & @UWStat student is learning from great Profs. Best part of being a Prof is learning from great students! Thanks @AnnaCNeufeld (@daniela_witten lab) for presenting your work on data thinning to our group!

English

6.2K

Anna Neufeld@AnnaCNeufeld·6 Nis

@wu_biostat @UWBiostat @UWStat @daniela_witten Thanks for inviting me!!!

English

156

Anna Neufeld@AnnaCNeufeld·21 Oca

@DarwinAwdWinner @daniela_witten @LucyGao We have been exploring this in the context of the negative binomial distribution in an upcoming paper about analyzing scRNA-seq data! We found surprisingly good performance with estimated overdispersion parameters, but there is definitely more to explore!

English

Ryan Thompson@DarwinAwdWinner·20 Oca

@AnnaCNeufeld @daniela_witten @LucyGao After reading it, the main thing I'm wondering is, what happens when you estimate nuisance parameters (like var for Gaussian) from the data itself rather than specifying them a priori? Presumably some kind of overfitting?

English

Anna Neufeld@AnnaCNeufeld·20 Oca

@daniela_witten, @lucygao, Ameer Dharamshi, and I are excited to share our new preprint (arxiv.org/abs/2301.07276). We introduce *data thinning*, a flexible framework that splits a single observation into independent parts, providing an alternative to cross-validation. (1/11)

English

108

27.3K

Anna Neufeld@AnnaCNeufeld·21 Oca

@DarwinAwdWinner @daniela_witten @LucyGao Great question! This is addressed in our previous paper, which was all about Poisson data thinning for scRNA-seq data! academic.oup.com/biostatistics/…

English

Ryan Thompson@DarwinAwdWinner·21 Oca

@AnnaCNeufeld @daniela_witten @LucyGao Also, since Poisson = NegBin w/0 disp, if the true dist is NegBin with >0 disp, then using Poisson will underestimate the var and result in >0 correlation between thinned data sets and overfitting, right?

English

Anna Neufeld@AnnaCNeufeld·20 Oca

To learn more about data thinning, check out our preprint, R package, and accompanying tutorial (anna-neufeld.github.io/datathin/). (11/11)

English

1.7K

Anna Neufeld@AnnaCNeufeld·20 Oca

Data thinning is broadly applicable: any time you might perform sample splitting or cross-validation, you can use data thinning instead. And you can use data thinning in settings where sample splitting and cross-validation are NOT applicable (e.g. unsupervised learning). (10/11)

English

1.6K

Anna Neufeld retweetledi

Daniela Witten@daniela_witten·26 Kas

And today a paper accepted in @biostatistics. Good week for @AnnaCNeufeld (and coauthors @lucylgao @popp_josh @alexisjbattle) 😁 🥳 🎉 👏

Daniela Witten@daniela_witten

Congrats to @AnnaCNeufeld for her first dissertation paper (with @lucylgao) being published in JMLR, and for winning the Birnbaum Award from @UWStat for an excellent general exam!! 🥳 🎉 👏 jmlr.org/papers/v23/21-…

English

Anna Neufeld@AnnaCNeufeld·26 Kas

Super excited about this paper!! And huge thanks to @daniela_witten for the ~extensive help~ with my general exam slides :)

Daniela Witten@daniela_witten

English

Keşfet

@MHemingNeuro @ScienceScottT @tangming2005 @hoffman_steven @Spottnik @UWBiostat @UWStat @daniela_witten