Mailman 3 August 2019 - scikit-learn

ANN: DIPY 1.0.0 - a historic release
by Eleftherios Garyfallidis Aug. 17, 2019

Aug. 17, 2019

We are excited to announce a new major and historic release of Diffusion Imaging in Python (DIPY). DIPY 1.0.0 is out! Please cite using the following DOI: 10.3389/fninf.2014.00008 <https://www.ncbi.nlm.nih.gov/pubmed/24600385> DIPY 1.0.0 (Monday, 5 August 2019) This release received contributions from 17 developers (the full release notes are at: https://dipy.org/documentation/1.0.0./release_notes/release1.0/). Thank you all for your contributions and feedback! A new DIPY era is starting: this release is compatible with python 3.5+ and breaks backward compatibility with 0.x.x. Please click here <https://dipy.org/documentation/1.0.0./api_changes/> to check API changes or look at the end of this email. The 0.16.x series will have extended bug-fix-only support for Python 2.7 until June 2020. Highlights of this release include: - Critical API changes <https://dipy.org/documentation/1.0.0./api_changes/> - New awesome website <http://dipy.org> - Large refactoring of tracking API - New denoising algorithm: MP-PCA - New Gibbs ringing removal - New interpolation module: dipy.core.interpolation - New reconstruction models: Mean Signal DKI, MTMS-CSD - Increased coordinate systems consistency - New object to manage safely tractography data: StatefulTractogram - New command line interface for downloading datasets: FetchFlow - Horizon updated, medical visualization interface powered by QuickBundlesX - Removed all deprecated functions and parameters - Removed compatibility with Python 2.7 - Updated minimum dependencies version (Numpy, Scipy) - All tutorials updated to API changes and 3 new added - Large documentation update - Closed 289 issues and merged 98 pull requests Note: - DIPY 0.16.x will be the last series to support python 2. The next release, DIPY 1.0, will support python 3 only. To upgrade or install <http://dipy.org/release0.10.html> DIPY Run the following command in your terminal: <http://dipy.org/release0.10.html> pip install --upgrade dipy or conda install -c conda-forge dipy This version of DIPY depends on nibabel (2.4.0+). For visualization you need FURY (0.3.0+). Questions or suggestions? For any questions go to http://dipy.org, or send an e-mail to dipy(a)python.org <neuroimaging(a)python.org> We also have an instant messaging service and chat room available at https://gitter.im/nipy/dipy On behalf of the DIPY developers, Eleftherios Garyfallidis, Ariel Rokem, Serge Koudoro https://dipy.org/contributors API Changes Some of the changes introduced in the 1.0 release will break backwards compatibility with previous versions. This release is compatible with Python 3.5+ Reconstruction The spherical harmonics bases mrtrix and fibernav have been renamed to tournier07 and descoteaux07 after the deprecation cycle started in the 0.15 release. We changed dipy.data.default_sphere from symmetric724 to repulsion724 which is more evenly distributed. Segmentation The API of dipy.segment.mask.median_otsu has changed in the following ways: if you are providing a 4D volume, vol_idx is now a required argument. The order of parameters has also changed. Tractogram loading and saving The API of dipy.io.streamlines.load_tractogram and dipy.io.streamlines.save_tractogram has changed in the following ways: When loading trk, tck, vtk, fib, or dpy) a reference nifti file is needed to guarantee proper spatial transformation handling. Spatial transformation handling Functions from dipy.tracking.streamlines were modified to enforce the affine parameter and uniform docstrings. deform_streamlines select_by_rois, orient_by_rois, _extract_vals and values_from_volume. Functions from dipy.tracking.utils were modified to enforce the affine parameter and uniform docstring. density_map connectivity_matrix, seeds_from_mask, random_seeds_from_mask, target, target_line_based, near_roi, length and path_length were all modified. The function affine_for_trackvis, move_streamlines, flexi_tvis_affine and get_flexi_tvis_affine were deleted. Functions from dipy.tracking.life were modified to enforce the affine parameter and uniform docstring. voxel2streamline, setup and fit from class FiberModel were all modified. afq_profile from dipy.stats.analysis was modified in a similar way. Simulations - dipy.sims.voxel.SingleTensor has been replaced by dipy.sims.voxel.single_tensor - dipy.sims.voxel.MultiTensor has been replaced by dipy.sims.voxel.multi_tensor - dipy.sims.voxel.SticksAndBall has been replaced by dipy.sims.voxel.sticks_and_ball Interpolation All interpolation functions have been moved to a new module name dipy.core.interpolation Tracking The voxel_size parameter has been removed from the following function: - dipy.tracking.utils.connectivity_matrix - dipy.tracking.utils.density_map - dipy.tracking.utils.stremline_mapping - dipy.tracking._util._mapping_to_voxel The dipy.reconst.peak_direction_getter.PeaksAndMetricsDirectionGetter has been renamed dipy.reconst.peak_direction_getter.EuDXDirectionGetter. The LocalTracking and ParticleFilteringTracking functions were moved from dipy.tracking.local.localtrackingto dipy.tracking.local_tracking. They now need to be imported from dipy.tracking.local_tracking. - functions argument tissue_classifier were renamed stopping_criterion The TissueClassifier were renamed StoppingCriterion and moved from dipy.tracking.local.tissue_classifier to dipy.tracking.stopping_criterion. They now need to be imported from dipy.tracking.stopping_criterion. - TissueClassifier -> StoppingCriterion - BinaryTissueClassifier -> BinaryStoppingCriterion - ThresholdTissueClassifier -> ThresholdStoppingCriterion - ConstrainedTissueClassifier -> AnatomicalStoppingCriterion - ActTissueClassifier -> ActStoppingCriterion - CmcTissueClassifier -> CmcStoppingCriterion The dipy.tracking.local.tissue_classifier.TissueClass was renamed dipy.tracking.stopping_criterion.StreamlineStatus. The EuDX tracking function has been removed. EuDX tractography can be performed using dipy.tracking.local_tracking using dipy.reconst.peak_direction_getter.EuDXDirectionGetter. Streamlines dipy.io.trackvis has been removed. Use dipy.io.streamline instead. Other - dipy.external package has been removed. - dipy.fixes package has been removed. - dipy.segment.quickbundes module has been removed. - dipy.reconst.peaks module has been removed. - Compatibility with Python 2.7 has been removed

1 0

Fwd: [Nairobi, Kenya WiMLDS] OS sprint (June 2019): Impact Report
by Reshama Shaikh Aug. 13, 2019

Aug. 13, 2019

Hello, On March 12, 2019, I had posted to this mailing list looking for a contributor for our Nairobi, WiMLDS sprint: https://mail.python.org/pipermail/scikit-learn/2019-March/003048.html Happy to share the sprint happened (on June 22, 2019) and I just published the Impact Report: https://reshamas.github.io/nairobi-wimlds-2019-scikit-learn-sprint-impact-r… Thanks to all contributors who assisted. Best, Reshama ---------------------------------------------- Reshama Shaikh <https://reshamas.github.io> NYC WiMLDS <https://meet.meetup.com/wf/click?upn=XTL-2BADyqX-2FD82kkVyMmL1UTawfruRN9Xsi…> NYC PyLadies <https://meet.meetup.com/wf/click?upn=pEEcc35imY7Cq0tG1vyTt6zEs68RbcMfjPcajN…> ----------------------------------------------

2 2

Along with Predictive probability and class name how to get the multi-text(text) belonging to that class
by Gunasekar K Aug. 13, 2019

Aug. 13, 2019

Hi Team, Greetings!!! Assuming ''Humpty Dumpty" is text trained under given specific model belonging to class called "cartoon character" in multi-text classification model. *model.predict_proba([Invoice_parameter])*100* gives the probability of belongings to specific class. But how come I can retrieve the *text* for that probability hit for that class. for example, Invoice_parameter= input() ---- on the fly input "Humpty Dumpty sat on the wall" *model.predict_proba([Invoice_parameter])*100* I get probability of 99.09 (rounded to two decimal places) to character class but i also need to retrieve the text Humpty Dumpty. Any help or suggestions are appreciated. Thanks in advance. Kind regards, GK

2 2

Using sklearn-crfsuite on Production Systems
by Astha Agarwal Aug. 6, 2019

Aug. 6, 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

Re: [scikit-learn] scikit-learn Digest, Vol 41, Issue 2
by Maiia Bakhova Aug. 5, 2019

Aug. 5, 2019

You probably can use LabelEncoder with OneHotEncoder from sklearn.preprocessing to convert your string data into numerical. Best, Mya On Thu, Aug 1, 2019 at 2:02 PM <scikit-learn-request(a)python.org> wrote: > Send scikit-learn mailing list submissions to > scikit-learn(a)python.org > > To subscribe or unsubscribe via the World Wide Web, visit > https://mail.python.org/mailman/listinfo/scikit-learn > or, via email, send a message with subject or body 'help' to > scikit-learn-request(a)python.org > > You can reach the person managing the list at > scikit-learn-owner(a)python.org > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of scikit-learn digest..." > > > Today's Topics: > > 1. SVM-RFE with scoring = 'f1' (Malik Yousef) > 2. Re: SVM-RFE with scoring = 'f1' (Guillaume Lema?tre) > 3. Re: SVM-RFE with scoring = 'f1' (Malik Yousef) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Thu, 1 Aug 2019 16:10:26 -0400 > From: Malik Yousef <malik.yousef(a)gmail.com> > To: scikit-learn(a)python.org > Subject: [scikit-learn] SVM-RFE with scoring = 'f1' > Message-ID: > <CACXKj+QQ9kZ-Q6PhwU+bQpxyBN-NOzmzjsBYCgW= > LvS+C7X_Zw(a)mail.gmail.com> > Content-Type: text/plain; charset="utf-8" > > Hello > When in using the scoring to be 'f1' then i get an error. > Here is the code and the error > > X=data > y=target_column > classifier = LinearSVC() > rfecv = RFECV(estimator=classifier, step=0.1, > cv=StratifiedKFold(5),scoring='f1') > rfecv.fit(X, y) > > The error is : > ValueError: pos_label=1 is not a valid label: array([u'c.pre', u'n.pre'], > dtype='<U5') > > Please your help > (I'm new to using scikit-leanr) > > --------------------------------------------------------------------------------------- > *Prof. Malik Yousef , Associate Professor * > *Zefat Academic College , Department of Information System * > Home Page: > https://malikyousef.com/ > Google Scholar Profile : > https://scholar.google.com/citations?user=9UCZ_q4AAAAJ&hl=en&oi=ao > > ---------------------------------------------------------------------------------------------------- >

1 0

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