🌐 Mailman 3 July 2019 - scikit-learn - python.org

Version 0.21! and plot_tree!
by Andrew Howe Dec. 5, 2019

Dec. 5, 2019

I want to say thank you to all the sklearn developers. The breadth and quality of this software is truly breathtaking. Specifically, I want to say thank you very very much for the plot_tree function! I have wasted a lot of effort in the past, on multiple OSes, getting everything to work so I could view the tree.export_graphviz results. Having this new function to plot the trees natively in matplotlib is extremely useful. Thanks again! Andrew <~~~~~~~~~~~~~~~~~~~~~~~~~~~> J. Andrew Howe, PhD LinkedIn Profile <http://www.linkedin.com/in/ahowe42> ResearchGate Profile <http://www.researchgate.net/profile/John_Howe12/> Open Researcher and Contributor ID (ORCID) <http://orcid.org/0000-0002-3553-1990> Github Profile <http://github.com/ahowe42> Personal Website <http://www.andrewhowe.com> I live to learn, so I can learn to live. - me <~~~~~~~~~~~~~~~~~~~~~~~~~~~>

2 4

Re: [scikit-learn] Monthly meetings between core developers
by Hanmin Qin Sept. 18, 2019

Sept. 18, 2019

I don't think it's worthwhile to worry too much about Beijing time if I'm the only person in Beijing. I'm happy to get up early once a month to learn from the great team! Hanmin Qin ----- Original Message ----- From: Alexandre Gramfort <alexandre.gramfort(a)inria.fr> To: Scikit-learn mailing list <scikit-learn(a)python.org> Subject: Re: [scikit-learn] Monthly meetings between core developers Date: 2019-07-18 17:16 hi, I kind of like the project boards we used for sprints: https://github.com/scikit-learn/scikit-learn/projects/11 the outcome of the core devs meeting could be to agree what should be listed on such a priority board. my 2c Alex On Thu, Jul 18, 2019 at 10:59 AM Olivier Grisel <olivier.grisel(a)ensta.org> wrote: > > I just found this planner to give it a try: > > https://www.timeanddate.com/worldclock/meetingtime.html?day=29&month=7&year… > > (Berlin and Paris are on the same timezone so I did not put only Berlin). > > It's going to be challenging to find a timeslot for every body. The > least extreme timeslot for everybody to attend at the same time would > be: > > https://www.timeanddate.com/worldclock/meetingdetails.html?year=2019&month=… > > We could also arrange for a second timeslot later (that would be > Tuesday morning in Australia and China): > > https://www.timeanddate.com/worldclock/meetingdetails.html?year=2019&month=… > > I wouldn't mind doing a meeting around 11pm on Monday evening from > time to time but it would still be very early for Beijing. > > Just to let you know, I will be off from next Saturday till Monday > August 19 (big summer break :) so don't count on my for the first > meeting if you start the meetings in the mean time. > > Le jeu. 18 juil. 2019 à 00:15, Andreas Mueller <t3kcit(a)gmail.com> a écrit : > > > > > > > > On 7/17/19 2:17 PM, Guillaume Lemaître wrote: > > > I am +1. This is a great initiative. > > > > > > IMO, we could make it really regular (i.e., a specific week-day of a > > > specific week in a month), with a rolling time (for the time-zone issue). > > > In this matter, we could maybe clear more in advance our agenda > > > instead of trying to find a date which accommodates everyone. > > > > > I agree, we could do something like the last Monday every month and > > alternate between two (or three) different time zones. > > We have CET (UTC+1), EST (UTC-5), CT (UTC+08), AEDT (USC+11) so that > > seems super easy, right? > > (TIL CST can stand for "Central"/US, China, and Cuba! not confusing at all) > > > > I agree that we should be as inclusive as possible, but I also don't > > want to create the expectation that some people (not thinking of any > > Australian in particular) > > who already sacrifice a lot of their free time have to invest even more > > time to keep up with the rest. > > > > I think the idea of posting write-ups will help being more inclusive in > > that regard. > > _______________________________________________ > > scikit-learn mailing list > > scikit-learn(a)python.org > > https://mail.python.org/mailman/listinfo/scikit-learn > > > > -- > Olivier > http://twitter.com/ogrisel - http://github.com/ogrisel > _______________________________________________ > scikit-learn mailing list > scikit-learn(a)python.org > https://mail.python.org/mailman/listinfo/scikit-learn _______________________________________________ scikit-learn mailing list scikit-learn(a)python.org https://mail.python.org/mailman/listinfo/scikit-learn

9 19

Using sklearn-crfsuite on Production Systems
by Astha Agarwal Aug. 5, 2019

Aug. 5, 2019

Hi, I'm wondering if anyone is using sklearn-crfsuite on production systems? Is this library suitable for usage in industry on production systems (and not academia) for non-big data problems? Thanks, Astha

3 2

cross-validated MANOVA
by charujing123 Aug. 1, 2019

Aug. 1, 2019

Dear experts and users, Does anyone know how to perform cross-validated multivariate analysis of variance? This is the paper mentioned this method "Searchlight-based multi-voxel pattern analysis of fMRI by cross-validated MANOVA". Thanks. Rujing 2019-07-31 charujing123

2 1

ANN Scikit-learn 0.21.3 and 0.20.4 released
by Joel Nothman July 30, 2019

July 30, 2019

We have released patches to Scikit-learn 0.21 (python >=3.5) and 0.20 (python 2 and 3) series including several bug fixes. See their respective change logs at https://scikit-learn.org/dev/whats_new/v0.21.html#version-0-21-3 and https://scikit-learn.org/dev/whats_new/v0.20.html#version-0-20-4. Install them from PyPI or conda-forge: * https://pypi.org/project/scikit-learn/0.21.3/ * https://pypi.org/project/scikit-learn/0.20.4/ Thanks to all who have contributed! Also, work in progress: Scikit-learn 0.22 is in development with lots of great new features to be released hopefully towards the end of 2019. See change logs at https://scikit-learn.org/dev/whats_new/v0.22.html for some of the things coming your way, and try them out by installing the nightly build (see https://scikit-learn.org/dev/developers/advanced_installation.html#installi… <https://scikit-learn.org/stable/developers/advanced_installation.html#insta…> ). Happy learning! the scikit-learn team

1 0

FINAL CALL for Papers: MICCAI 2019 Connectomics in NeuroImaging Workshop and Challenge
by Ai Wern July 25, 2019

July 25, 2019

**Apologies for cross posting** Dear Colleagues, This is a final call for full-length papers submissions to our *3rd International Workshop on Connectomics in NeuroImaging (CNI 2019), *and *Transfer-Learning CNI Challenge 2019 *held in parallel with the 22nd International Conference on Medical Image Computing and Computer-assisted Intervention (MICCAI 2019) in Shenzhen, China. *Our deadline has been extended to Weds 31st July 2019* **** CNI Workshop Call for Papers **** Our topics of interest cover (but are not limited to): (1) New developments in connectome construction from different imaging modalities; (2) Development of data driven techniques to identify biomarkers in connectome data; (3) Machine learning algorithms and connectome data analysis; (4) Brain network modeling and formal conceptual models of connectome data; (5) Evaluation and validation of connectome models. If you have research that fits into the scope of our workshop detailed on our website (http://www.brainconnectivity.net/workshop*)*, we encourage you to submit a paper. **** CNI Call for Challengers **** Addressing the issues of generalizability and clinical relevance for functional connectomes, you can leverage a unique resting-state fMRI (rsfMRI) dataset of attention deficit hyperactivity disorder (ADHD) and neurotypical controls (NC) to design a classification framework that can predict subject diagnosis (ADHD vs. NC) based on brain connectivity data. In a surprise twist, we will also evaluate the classification performance on a related clinical population with an ADHD comorbidity. This challenge will allow us to assess (1) whether the method is extracting functional connectivity patterns related to ADHD symptomatology, and (2) how much of this information “transfers” between clinical populations. Training and validation data are now released http://www.brainconnectivity.net/challenge **** Why submit to the CNI Workshop and Challenge? **** - Keynote talks by Prof Yong He (Beijing Normal University, China) and Dr. Fan Zhang (Harvard Medical School, USA); - Oral presentations and poster sessions to provide you with ample opportunity for exchanges and discussions; - Accepted papers will be published in an LNCS proceedings; - Best Paper and Poster Awards will be presented, and sponsored prizes for Challenge winners. **** Important dates for CNI workshop **** - Submission deadline: July 31st, 2019, 23:59 EST - Notification of acceptance: August 13th, 2019 - Camera-ready deadline : August 18th, 2019, 23:59 EST - Submission website: https://cmt3.research.microsoft.com/CNI2019 **** Important dates for CNI Challenge **** - Submission deadline: August 15th, 2019, 23:59 EST - Submission website: https://cmt3.research.microsoft.com/CNIChallenge2019 <https://cmt3.research.microsoft.com/CNI2019> For more information, visit our website at http://www.brainconnectivity.net. Best wishes, CNI 2019 Organising Committee

1 0

question using GridSearchCV
by Glenn Schultz July 24, 2019

July 24, 2019

I am using GBClassifier, the below works if I use the default accuracy but it fails using roc_auc or roc_auc_score. I have found many examples to work with but for the life of me I can’t get it two work with roc_auc. What am I doing wrong. from sklearn.ensemble import GradientBoostingClassifier from sklearn.model_selection import GridSearchCV from sklearn.metrics import auction, roc_auc_score y_train = LoansTrainData[‘event’] x_train LoanTrainData[LoansTrainData.columns.drop(‘event’)] parameters = {“loss” :[‘deviance’], “scoring” :[‘roc_auc’}, “learning_rate” :[.1, .05] selfCLF =GridSearchCV(GradientBoostingClassifier(), parameters, versose = 3m n_jobs = 4) searchCLF(x_train, y_train)

2 2

Random Forest without target to measure feature importance
by Gabor Toth July 23, 2019

July 23, 2019

Hello, I would like to use Random Forest classifier to assess the importance of features (bag-of-words) but I don't have any predefined class labels or any test data. I have earlier used ExtraTreesClassifier() with fit_transform, which is not available anymore (see below). I am wondering how I could use Random Forest now. clf = ExtraTreesClassifier() clf.fit_transform(doc_term_matrix,np.empty(doc_term_matrix.shape)) features_importance=np.array(clf.feature_importances_) Thanks, Gabor

1 0

Long term roadmap and moonshot goals
by Andreas Mueller July 23, 2019

July 23, 2019

Hi all. At SciPy, Brian Granger raised a good point about their planning for the Jupyter Project, which is the importance of long-term goals. I think it's great that we now have a detailed short-term roadmap (https://scikit-learn.org/dev/roadmap.html). Given that we now have about 6(!) full time people (Oliver, Jeremy, Guillaume, Nicolas, Thomas, Adrin) on scikit-learn (GO TEAM!!), I think it's realistic to achieve most of these within a year or two. We have actually made some significant progress already. I think now would be a good time to start thinking about a longer-term roadmap, say 3-5 years out. What do we want to achieve? What are realistic goals, and what are moonshot goals? Having a common vision and shared goals might help us with funding, but might also help us with prioritization and motivation. What do you think? Do you think this is important and worth-while? And what should our goals be? Best, Andy

4 5

Continues monitoring of benchmark performances
by Adrin July 23, 2019

July 23, 2019

Hi, There is this [page](https://pandas.pydata.org/speed/scikit-learn/) maintained by some of the pandas maintainers (@TomAugspurger in particular), and it seems like a really good idea to have an eye on the performance of different benchmarks through time just in case a PR introduces some major drawbacks. However, he doesn't have the bandwidth to maintain it much more, and not really the hardware. I think it'd be a good idea for us to have that, wanted to bring it up and see what you think! Cheers, Adrin.

6 5