Libraries for Online Machine Learning

Question

I am looking for packages (either in python, R, or a standalone package) to perform online learning to predict stock data.

I have found and read about Vowpal Wabbit (https://github.com/JohnLangford/vowpal_wabbit/wiki), which seems to be quite promising but I am wondering if there are any other packages out there.

Thanks in advance.

Dumb question, but do you mean `online` as in "non-batch mode", or as in "processed in the cloud"? — Mike Ounsworth, Sep 09 '14 at 17:14
Hi Mike, I mean non-batch mode. So soon after a prediction is made we learn the true label and then use that in the training. So think in terms of predicting a stock price, in a few minutes we would learn the true value, and then use that value in our training. — mike1886, Sep 11 '14 at 13:16
Not online learning, but take a look at this post, this may help you to get a start http://francescopochetti.com/stock-market-prediction-part-introduction/ — Adriano Almeida, Oct 09 '14 at 10:22

score 3 · Answer 1 · edited Jan 23 '16 at 22:19

3

You could look at scikit-learn or Orange module in Python. Scikit-learn has a SGD classifier and regressor that could do a partial fit data in case of online learning. In R, take a look at caret package

edited Jan 23 '16 at 22:19

K3---rnc

3,442
1
12
12

answered Sep 05 '14 at 06:17

Rahul Agarwal

191
3

Thanks for your comments. I am very aware and use scikit and caret regularly but I am looking for more of a state of the art online learning package, and neither scikit or caret are these. I am looking for people who are actually using something and can give their experience with a given package. – mike1886 Sep 05 '14 at 13:23
1

@mike1886: if you already know how to update your model using new data, what else do you want? What is "more of state of the art" in your understanding? Just more recent and sounding cooler? – ffriend Oct 06 '14 at 05:56
Agree---scikit's online learning modules are pretty 'state-of-the-art', whatever that means – Ben Allison Oct 13 '14 at 13:56
@Rahul in R is caret package supports any online learning algorithm? if yes, please suggest. – Divyang Shah Jun 09 '17 at 10:01

score 1 · Answer 2 · answered Sep 17 '19 at 11:14

There are a few other libraries for online or incremental machine learning.

Jubatus, which is written in C++, but it officially provides a C++, Python, Ruby and Java clients.
The Tornado Framework, written in Python
creme, written in Python
LIBOL, written in C++ (and Matlab). You can find a more detailed description of the library in the paper LIBOL: A Library for Online Learning Algorithms (2014)
LibTopoART, written in C#
scikit-learn, which provides some online and incremental learning functionalities

OliverHennhoefer · Answer 3 · 2022-09-20T18:13:11.997

0

River - Online Machine Learning in Python (website) previously creme and scikit-multiflow.

edited Sep 20 '22 at 18:13

answered Sep 20 '22 at 12:09

OliverHennhoefer

328
2
3
11

Libraries for Online Machine Learning

3 Answers3