Mailman 3 February 2020 - scikit-learn

ANN: scikit-learn 0.22 final release
by Adrin March 4, 2020

March 4, 2020

We're happy to announce the 0.22 release. You can read the release highlights under https://scikit-learn.org/stable/auto_examples/release_highlights/plot_relea… and the long version of the change log under https://scikit-learn.org/stable/whats_new/v0.22.html#changes-0-22. This version supports Python versions 3.5 to 3.8. You can give it a go using `pip install -U scikit-learn` while conda and conda forge binaries are coming. Regards, Adrin, on behalf of the scikit-learn maintainer team.

5 6

Conference call regarding feature names
by Adrin Feb. 24, 2020

Feb. 24, 2020

Hi, We agreed to have a conference call on March 9th, to talk about feature names issues, pandas in, pandas out, xarray, etc. Time/Date: https://www.timeanddate.com/worldclock/meetingdetails.html?year=2020&month=… Here's the info to the call: Join Zoom Meeting https://anaconda.zoom.us/j/158272910?pwd=NURzbWhTSXNYNFhRMVYxRFVybjUrdz09 Meeting ID: 158 272 910 Password: 5867453 Find your local number: https://anaconda.zoom.us/u/adk50yIacN Best, Adrin.

1 0

Monthly meetings
by Nicolas Hug Feb. 24, 2020

Feb. 24, 2020

Hi all, The next scikit-learn monthly meeting will take place on Monday at the usual time (https://www.timeanddate.com/worldclock/meetingdetails.html?year=2020&month=… <https://www.google.com/url?q=https://www.timeanddate.com/worldclock/meeting…> While these meetings are mainly for core-devs to discuss the current topics, we're also happy to welcome non-core devs and other projects maintainers! Feel free to join. *Location has changed:* Join Zoom Meeting https://anaconda.zoom.us/j/947129165?pwd=dEFZNHM0ZFBiQWlDYlJlRW1EaHg2QT09 <https://www.google.com/url?q=https://anaconda.zoom.us/j/947129165?pwd%3DdEF…> Meeting ID: 947 129 165 Password: 586745 *@core devs, please make sure to update your notes on Friday* Thanks, Nicolas

4 4

Nonlinear partial least square (PLS)
by Andrea Moglia Feb. 23, 2020

Feb. 23, 2020

Hi, I need to use nonlinear partial least square (PLS), such as kernel PLS. Do you know how to do it in Scikit Learn? Thank you in advance. Andrea

1 0

Domain adaptation and cross-validation
by Johan Mazel Feb. 20, 2020

Feb. 20, 2020

Hello I am working on a binary classification task using machine learning. One class C0 is built upon public data called D0. The other class C1 is made of data named D1 that I generated. I actually generated two datasets D1a and D1b with slightly different parameters. I would like to evaluate the domain adaptation of a model in the context of D0, D1a and D1b. It means that I would like to train a model on data from D0 and D1a, and then, test its performance on D0 and D1b. I plan to perform a k-fold cross-validation on D0 and bootstrapping on D1a on D1b. For example, at each iteration, I would build training data from k-1 folds of D0 and boostrapped data from D1a. Testing data would be built upon the single remaining fold of D0 and bootstrapped data from D1b. I was thinking of using a class similar to those in sklearn.model-selection (e.g. KFold or StratifiedKFold) to perform the method described above. The init function of the class would be initialized with the KFold/StratifiedKFold and bootstrapping parameters. The split function inside this class would be generic enough to handle many datasets for either D0, or D1a, or D1b. This function's parameters would be the usual X and y for data and targets, along with information about the structure of X. This additional information would be propagated from the fit function in a similar way as optional groups parameter is passed to group-related split functions in classes such as GroupKFold. Here X would be the concatenation of train-test datasets (i.e. D0-like), train only datasets (i.e. D1a-like) and test only datasets (i.e. D1b-like). y would be built in a similar way. The additional parameters may thus be two tuples for the start and end indexes of train only datasets (i.e. D1a-like) and test only datasets (i.e. D1b-like). These values would allow the split function to properly operate on X and y by taking into account boundaries between dataset types when building folds and performing bootstrapping. As far as I know, I cannot perform such a procedure with scikit-learn using the functions in sklearn.model-selection (https://github.com/scikit-learn/scikit-learn/blob/master/sklearn/model_sele…). Did I miss something? Maybe somewhere else in the code? If there is no implementation in scikit-learn, would you be interested in a pull-request with such a function? Best regards, Johan Mazel Les données à caractère personnel recueillies et traitées dans le cadre de cet échange, le sont à seule fin d’exécution d’une relation professionnelle et s’opèrent dans cette seule finalité et pour la durée nécessaire à cette relation. Si vous souhaitez faire usage de vos droits de consultation, de rectification et de suppression de vos données, veuillez contacter contact.rgpd(a)sgdsn.gouv.fr. Si vous avez reçu ce message par erreur, nous vous remercions d’en informer l’expéditeur et de détruire le message. The personal data collected and processed during this exchange aims solely at completing a business relationship and is limited to the necessary duration of that relationship. If you wish to use your rights of consultation, rectification and deletion of your data, please contact: contact.rgpd(a)sgdsn.gouv.fr. If you have received this message in error, we thank you for informing the sender and destroying the message.

1 0

Need help getting started with raster pattern learning
by Stephen Woodbridge Feb. 19, 2020

Feb. 19, 2020

Hi all, I think I need some help with getting started to use scikit for a project. Having a basic strategy would be helpful. I've been reading the docs and examples and can see how various tools might apply, but more questions than answers at the moment. So my goal is to take various observations at locations in the ocean and then using satellite data at those locations, use that as training data. Then later evaluate new satellite raster images to identify similar locations in the new images. So given a location, I plan to extract a region around that location of some size from the satellite data because the area around the point is likely significant, not just the point. So say, I have 500x500 extracted image. I think I want to use a one-class SVM because I only have positive samples and no negative samples, but I'm not sure how to transform the extracted raster into something that can be fed into the SVM for training. Likewise, what if I had multiple extracted images, like an SST and a Chlorophyll, for a given location. Could this be handled as a multiple band image? Later given a trained SVM, what would I use to evaluate a larger satellite image to find simial locations. Not ask for a solution, but what tools should I be looking at for the different steps in this process. Thanks, -Steve

1 0

SLEP013 VOTE: n_features_out_
by Adrin Feb. 19, 2020

Feb. 19, 2020

Hi, SLEP013 <https://scikit-learn-enhancement-proposals.readthedocs.io/en/latest/slep013…> proposes to add an "n_features_out_" attribute to track the number of output features to all estimators. Please cast your vote on this PR <https://github.com/scikit-learn/enhancement_proposals/pull/36> which proposes to accept the SLEP. Final comments are more than welcome. Following is the short version of the SLEP. Regards, Adrin. Abstract ######## This SLEP proposes the introduction of a public ``n_features_out_`` attribute for most transformers (where relevant). Motivation ########## Knowing the number of features that a transformer outputs is useful for inspection purposes. This is in conjunction with `*SLEP010: ``n_features_in_``* < https://scikit-learn-enhancement-proposals.readthedocs.io/en/latest/slep010… >`_. Solution ######## The proposed solution is for the ``n_features_out_`` attribute to be set once a call to ``fit`` is done. In many cases the value of ``n_features_out_`` is the same as some other attribute stored in the transformer, *e.g.* ``n_components_``, and in these cases a ``Mixin`` such as a ``ComponentsMixin`` can delegate ``n_features_out_`` to those attributes.

1 0

Need for multioutput multivariate algorithm for Random Forest in Python (using Mahalanobis distance)
by Paul Chike Ofoche Feb. 17, 2020

Feb. 17, 2020

Hello all, My name is Paul and I am enthused about data science. I have been using Python and other programming languages for close to two years. There is an issue that I have been facing since I began applying Python to the analysis of my research work. My question has remained unanswered for months. Has anybody not run into the need to work with data whereby the regression results are a multiple output, in which the output parameters are correlated with each other? This is called a multi-output multivariate problem. A version of random forest that handles multiple outputs is referred to as the multivariate random forest. It is implemented in the programming language, R (see attached reference documentation below). Till date, there exists no such package in Python. My question is whether anybody knows how to go about implementing this. The random forest univariate regression case utilizes the Euclidean distance as the measurement criteria, whereas the multivariate regression case uses the Mahalanobis distance, which takes into account the inter-relationships between the multiple outputs. I have inquired about an equivalent capability in Python for many years, but it has still not been addressed. Such a multivariate random forest mode is very applicable to the type of research and analysis that I do. Could someone help, please? Thank you, Paul Ofoche PS: This is an important need for multivariate output analysis as a technique to solving practical research problems. Here are some posted questions by various other Python users concerning this same issue. https://datascience.stackexchange.com/questions/21637/code-for-multivariate… Multi-output regression | | | | | | | | | | | Multi-output regression I have been looking in to Multi-output regression the last view weeks. I am working with the scikit learn packag... | | |

4 7

ANN: PyDataLondon 2020 Call for Proposals (closing soon on February 21)
by Ian Ozsvald Feb. 12, 2020

Feb. 12, 2020

PyDataLondon 2020 runs on May 15-17, we'll have the same 3 day mult track format as last year [1] with circa 700 attendees. We're keen to have new talks and tutorials on machine learning, data science, data engineering and associated areas. We welcome and encourage first-time speakers, mentorship is available for first-time submitters to help refine your proposal: https://pydata.org/london2020/cfp/ Topics around scikit-learn, related scikit packages and related packages (e.g. PyMC3, Keras) are all very welcome. Thanks to NumFOCUS there are some travel and diversity grants. Our Call for Proposals is open until February 21, all proposals will be released to the double-blind review committee simultaneously at this point. This is our second year with a double-blind committee. In addition to a double-blind review process the conference offers a creche, prayer and quiet rooms and an "unconference" track which is scheduled by participants to increase the diversity of talks and activities, in addition to the usual pre-scheduled tracks: https://pydata.org/london2020/ If you have questions feel free to email me directly and I'll try to help. I'm not on the review committee, I'm one of the founders of the conference series and a continuing volunteer. All of the core PyDataLondon organisers are volunteers. Thanks, Ian. [1] https://pydata.org/london2019/schedule/ -- Ian Ozsvald (Data Scientist, PyDataLondon co-chair) ian(a)IanOzsvald.com https://IanOzsvald.com https://MorConsulting.com https://twitter.com/IanOzsvald

1 0

The exact formula used to compute the tf-idf
by Peng Yu Feb. 1, 2020

Feb. 1, 2020

Hi, I am trying to understand the exact formula for tf-idf. vectorizer = TfidfVectorizer(ngram_range = (1, 1), norm = None) wordtfidf = vectorizer.fit_transform(texts) Given the following 3 documents (id1, id2, id3 are the IDs of the three documents). id1 AA BB BB CC CC CC id2 AA AA AA AA BB BB BB BB BB DD DD DD DD DD DD id3 AA AA AA AA AA AA AA DD DD DD DD DD DD DD DD FF FF FF FF FF FF FF FF FF The results are the following. id1▸ cc▸ 5.079441541679836¬ id1▸ bb▸ 2.5753641449035616¬ id1▸ aa▸ 1.0¬ id2▸ dd▸ 7.726092434710685¬ id2▸ bb▸ 6.438410362258904¬ id2▸ aa▸ 4.0¬ id3▸ ff▸ 15.238324625039509¬ id3▸ dd▸ 10.301456579614246¬ id3▸ aa▸ 7.0¬ According to "6.2.3.4. Tf–idf term weighting" on the following page. https://scikit-learn.org/stable/modules/feature_extraction.html#text-featur… For aa, as n = 3 and df =3, idf(aa) = log((1+n)/(1+df)) + 1 = 1. But I don't understand why tf-idf(id1, aa) is 1. This means that tf(id1, aa) is 1, which is just the count of aa, shouldn't it be divided by the number of terms in the doc id1, which should result in 1/6 instead of 1? Thanks. -- Regards, Peng

2 1