Mailman 3 November 2020 - scikit-learn

Question RE: skLearn Logistic Regression
by The Helmbolds March 6, 2021

March 6, 2021

I have a case with binary results and 1-D features, like: X = np.array(-3,-2,-1,0,1,2,3,) and y = np.array(0, 0, 0, 1, 1, 1, 1) only longer arrays (about 180 entries in each array) of this general type. So this should be the "simplest" case. Altho I've tried several variations of the Logistic input formats, in LogisticRegression.fit(X, y) they keep being rejected with the most common error message being Missing argument y I assure you I do indeed have an array "y" that is passed to "fit" So----What do I have to do to get Logistic Regression to accept 1-D features?

3 2

Changes in Travis billing
by Gael Varoquaux Dec. 8, 2020

Dec. 8, 2020

Travis is changing it's billing strategy: https://blog.travis-ci.com/2020-11-02-travis-ci-new-billing Open repositories are getting a free initial set of credit. They invite open source projects to contact them to benefit from a more liberal policy. I suggest that we do the latter, as I fear that we might run out of credits, and I am quite convinced that we could benefit from the liberal policy. Cheers, Gaël -- Gael Varoquaux Research Director, INRIA Visiting professor, McGill http://gael-varoquaux.info http://twitter.com/GaelVaroquaux

8 12

Monthly meeting November 30th 2020
by Chiara Marmo Nov. 25, 2020

Nov. 25, 2020

Dear list, The next scikit-learn monthly meeting will take place on Monday October 30th at 8PM UTC: https://www.timeanddate.com/worldclock/meetingdetails.html?year=2020&month=… While these meetings are mainly for core-devs to discuss the current topics, we are also happy to welcome non-core devs and other project maintainers. Feel free to join, using the following link: https://meet.google.com/xhq-yoga-rtf If you plan to attend and you would like to discuss something specific about your contribution please add your name (or github pseudo) in the " Contributors <https://hackmd.io/1yDGruklTNqtx1Eb_uEP0Q#Contributors>" section, of the public pad: https://hackmd.io/1yDGruklTNqtx1Eb_uEP0Q Best Chiara

1 0

imbalanced datasets return uncalibrated predictions - why?
by Sole Galli Nov. 19, 2020

Nov. 19, 2020

Hello team, I am trying to understand why does logistic regression return uncalibrated probabilities with values tending to low probabilities for the positive (rare) cases, when trained on an imbalanced dataset. I've read a number of articles, all seem to agree that this is the case, many show empirical proof, but no mathematical demo. When I test it myself, I can see that this is indeed the case, Logit on imbalanced datasets returns uncalibrated probs. And I understand that it has to do with the cost function, because if we re-balance the dataset with say class_weight = 'balance'. then the probabilities seem to be calibrated as a result. I was wondering if any of you knows the mathematical demo that supports this conclusion? Any mathematical demo, or clear explanation of why logit would return uncalibrated probs when trained on an imbalanced dataset? Any link to a relevant article, video, presentation, etc, will be greatly appreciated. Thanks a lot! Sole

3 3

Incremental learning in scikit-learn
by Farzana Anowar Nov. 15, 2020

Nov. 15, 2020

Hello everyone, Currently, I am working with incremental learning. I know that scikit-learn allows using incremental learning for some classifiers i. e. SGD. In incremental learning, data is not available all together rather the data become available chunk by chunk over the time. Now, my question is: does scikit-learn allows to have different data chunk or all the chunks has to be of the same size? Thanks! -- Best Regards, Farzana Anowar, PhD Candidate Department of Computer Science University of Regina

2 1

Creating dataset
by Mahmood Naderan Nov. 8, 2020

Nov. 8, 2020

Hi, I have created an input file similar to iris data set. That is something like this: 0.1,0.2,0.3,0.4,M1 ... I want to know how I can create my own dataset similar to the following lines? from sklearn.datasets import load_iris iris = load_iris() Regards, Mahmood

5 8

implementing regularized random forest
by Mick Men Nov. 3, 2020

Nov. 3, 2020

Hello, I am trying to implement my own regularized random forest (RRF) which grows trees in series and selects new features only if they are better than the features used in previous splits. This is for a research project and I will need to ship the code with the publication. So far I have a working proof of concept where I modified the scikit-learn forest, tree, and splitter modules. But this mean that I need to ship my fork version of scikit-learn. Ideally, I am looking for a way to build my own RRF that uses scikit-learn API instead of modifying it. Is it possible? Thanks. Mickael

2 1

Issue with Sklearn.Logistic Regression
by The Helmbolds Nov. 1, 2020

Nov. 1, 2020

Here's my ynp and Xnp arrays: Print ynp [0 1 0 0 0 0 0 0 0 0 0 0 0 1 0 1 0 1 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 1 1 0 0 1 1 0 0 1 0 0 0 0 0 0 1 0 0 0 1 1 0 1 1 0 1 0 0 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 1 0 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1 0 0 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1] Shape of ynp = 160 Print Xnp [-3.00000000e+00 -2.95000000e+00 -2.90000000e+00 -2.85000000e+00 -2.80000000e+00 -2.75000000e+00 -2.70000000e+00 -2.65000000e+00 -2.60000000e+00 -2.55000000e+00 -2.50000000e+00 -2.45000000e+00 -2.40000000e+00 -2.35000000e+00 -2.30000000e+00 -2.25000000e+00 -2.20000000e+00 -2.15000000e+00 -2.10000000e+00 -2.05000000e+00 -2.00000000e+00 -1.95000000e+00 -1.90000000e+00 -1.85000000e+00 -1.80000000e+00 -1.75000000e+00 -1.70000000e+00 -1.65000000e+00 -1.60000000e+00 -1.55000000e+00 -1.50000000e+00 -1.45000000e+00 -1.40000000e+00 -1.35000000e+00 -1.30000000e+00 -1.25000000e+00 -1.20000000e+00 -1.15000000e+00 -1.10000000e+00 -1.05000000e+00 -1.00000000e+00 -9.50000000e-01 -9.00000000e-01 -8.50000000e-01 -8.00000000e-01 -7.50000000e-01 -7.00000000e-01 -6.50000000e-01 -6.00000000e-01 -5.50000000e-01 -5.00000000e-01 -4.50000000e-01 -4.00000000e-01 -3.50000000e-01 -3.00000000e-01 -2.50000000e-01 -2.00000000e-01 -1.50000000e-01 -1.00000000e-01 -5.00000000e-02 -2.28983499e-15 5.00000000e-02 1.00000000e-01 1.50000000e-01 2.00000000e-01 2.50000000e-01 3.00000000e-01 3.50000000e-01 4.00000000e-01 4.50000000e-01 5.00000000e-01 5.50000000e-01 6.00000000e-01 6.50000000e-01 7.00000000e-01 7.50000000e-01 8.00000000e-01 8.50000000e-01 9.00000000e-01 9.50000000e-01 1.00000000e+00 1.05000000e+00 1.10000000e+00 1.15000000e+00 1.20000000e+00 1.25000000e+00 1.30000000e+00 1.35000000e+00 1.40000000e+00 1.45000000e+00 1.50000000e+00 1.55000000e+00 1.60000000e+00 1.65000000e+00 1.70000000e+00 1.75000000e+00 1.80000000e+00 1.85000000e+00 1.90000000e+00 1.95000000e+00 2.00000000e+00 2.05000000e+00 2.10000000e+00 2.15000000e+00 2.20000000e+00 2.25000000e+00 2.30000000e+00 2.35000000e+00 2.40000000e+00 2.45000000e+00 2.50000000e+00 2.55000000e+00 2.60000000e+00 2.65000000e+00 2.70000000e+00 2.75000000e+00 2.80000000e+00 2.85000000e+00 2.90000000e+00 2.95000000e+00 3.00000000e+00 3.05000000e+00 3.10000000e+00 3.15000000e+00 3.20000000e+00 3.25000000e+00 3.30000000e+00 3.35000000e+00 3.40000000e+00 3.45000000e+00 3.50000000e+00 3.55000000e+00 3.60000000e+00 3.65000000e+00 3.70000000e+00 3.75000000e+00 3.80000000e+00 3.85000000e+00 3.90000000e+00 3.95000000e+00 4.00000000e+00 4.05000000e+00 4.10000000e+00 4.15000000e+00 4.20000000e+00 4.25000000e+00 4.30000000e+00 4.35000000e+00 4.40000000e+00 4.45000000e+00 4.50000000e+00 4.55000000e+00 4.60000000e+00 4.65000000e+00 4.70000000e+00 4.75000000e+00 4.80000000e+00 4.85000000e+00 4.90000000e+00 4.95000000e+00] Shape of Xnp = 160 Press ENTER to continue = Now Call Logistic Regression --------------------------------------------------------------------------- TypeError Traceback (most recent call last) <ipython-input-3-300811cacc0b> in <module> 103 104 aprint("Now Call Logistic Regression") --> 105 logreg = LogisticRegression.fit(Xnp, ynp) 106 aprint("Print logreg output") 107 print(logreg) TypeError: fit() missing 1 required positional argument: 'y' Eh!?!? What happened??? "You won't find the right answers if you don't ask the right questions!" (Robert Helmbold, 2013)

4 6