2015-07-11 22:43 GMT+02:00 Andreas Mueller t3k...@gmail.com:
It's a bit odd because numpy is not linked against iomp5. What is that /
It's the OpenMP runtime. And by the looks of it, MKL is linking with it.
--
Don't
2015-07-10 15:20 GMT+02:00 Al alain.pen...@gmail.com:
My name is Alain Pena, (now previously) student in computer engineering
at University of Liège.
For my master thesis, I had to implement some methods for multilabel
classification, those methods being RAKEL [1] and (Ensemble) Classifier
2015-07-01 16:27 GMT+02:00 Fred Mailhot fred.mail...@gmail.com:
2) The gensim implementation predates the patenting
Does that matter?
--
Don't Limit Your Business. Reach for the Cloud.
GigeNET's Cloud Solutions provide
2015-05-22 9:01 GMT+02:00 zhenjiang zech xu zhenjiang...@gmail.com:
I tested the following code and its outputs show predict_proba and predict
give very different result, even for the samples with high probability (0.7)
to be label 1 are predicted as label 1. I am very surprised. Is this
2015-05-22 8:29 GMT+02:00 Sebastian Raschka se.rasc...@gmail.com:
The default equation is:
# idf = log ( number_of_docs / number_of_docs_where_term_appears )
And in the online documentation at
2015-02-17 11:25 GMT+01:00 abhishek abhish...@gmail.com:
4294901761:21 4294902016:18 4294967041:15 4294967296:54
I am unable to understand why should it fail when maxint for python is
9223372036854775807.
Is there any workaround available for this? Or is it just not possible to
load at
2014-12-18 20:14 GMT+01:00 David Brough david.brough.0...@gmail.com:
Thank you for the quick response.
I currently using version 0.15.2. Our arrays have dimensions [n_samples, x,
y, z]. Below are the two trace-backs I get for both train_test_split and
GrindSearchCV. They both end at the same
2014-12-18 21:45 GMT+01:00 Alexandre Gramfort
alexandre.gramf...@telecom-paristech.fr:
there is an allow_nd param I added to support this use case. Maybe
we could add it to GridSearchCV...
We decided against this in the case of k-NN, where the suggestion (I
don't remember by whom) was to alllow
2014-12-16 17:57 GMT+01:00 Sebastian Raschka se.rasc...@gmail.com:
Maybe it's something in scipy according to Manoj's linked discussion ... in
any case, maybe a workaround for this issue and future issues would be to
have a
forxe_clear_gc (default=False) parameter to force the garbage
2014-12-04 0:55 GMT+01:00 Joel Nothman joel.noth...@gmail.com:
For example, let's say someone has implemented an algorithm (Affinity
Propagation is what triggered this discussion so you might consider that).
Someone else wants to come and add features to it, or even just clean the
code, but by
2014-11-27 17:26 GMT+01:00 Ian Ozsvald i...@ianozsvald.com:
If safe_sparse_dot is called with dense_output=False then I get a sparse
result and everything looks sensible with low RAM usage.
I'm using 0.15, the current github shows the line:
2014-11-18 11:07 GMT+01:00 Nicola Sambin sam...@spaziodati.eu:
- when I computed:
for vector in vectors:
classifier.predict_proba(vector)
it took:
2227,99s user 90,75s system 21:29,94 total
- while
classifier.predict_proba(vectors)
took:
1,06s user 0,39s system 1,984 total
What is
2014-11-18 16:38 GMT+01:00 Chris Holdgraf choldg...@berkeley.edu:
Sometimes this works, but sometimes it doesn't. For example, if I use
LinearSVC as my model, then setting .coef_ works fine. However, if I use
SVC(kernel='linear'), then I get an attribute error when I try to set any
attribute.
2014-11-02 22:09 GMT+01:00 Andy t3k...@gmail.com:
No. That would be backward stepwise selection. Neither that, nor its
forward cousin (find most discriminative feature, then second-most,
etc.) are implemented in scikit-learn.
Isn't RFE the backward step selection using a maximum number of
2014-10-23 16:21 GMT+02:00 George Bezerra gbeze...@gmail.com:
Is there a simple way to get the probabilities that a data point belongs to
a class for this model?
SVMs aren't probability models. You can use LogisticRegression, that's
the same algorithm but with a different loss function.
2014-10-21 4:14 GMT+02:00 Joel Nothman joel.noth...@gmail.com:
I assume Robert's query is about RFECV.
Oh wait, RFE = backward subset selection. I'm an idiot, sorry.
--
Comprehensive Server Monitoring with Site24x7.
2014-10-20 22:08 GMT+02:00 George Bezerra gbeze...@gmail.com:
Not an expert, but I think the idea is that you remove (or add) features one
by one, starting from the ones that have the least (or most) impact.
E.g., try removing a feature, if performance improves, keep it that way and
move on
2014-10-14 21:53 GMT+02:00 Robert Layton robertlay...@gmail.com:
Currently the word metrics is overloaded with at least two type of
algorithms in that module. The first is evaluation metrics and the second is
functions dealing with distance metrics.
My suggestion is to:
1) Move the
2014-09-09 3:36 GMT+02:00 Apu Mishra apumishra...@gmail.com:
Lars Buitinck larsmans@... writes:
The way to combine HV and
Tfidf is
hashing = HashingVectorizer(non_negative=True, norm=None)
tfidf = TfidfTransformer()
hashing_tfidf = Pipeline([(hashing, hashing), (tidf, tfidf)])
I notice
2014-10-08 11:32 GMT+02:00 Karimkhan Pathan karimkhan...@gmail.com:
can I use this trained file with nltk to classify `plain input text`?
This is the scikit-learn mailing list. You should be asking on the NLTK ML.
--
2014-10-07 23:03 GMT+02:00 Pagliari, Roberto rpagli...@appcomsci.com:
Do I just use the bug tracker?
You can, but we'd much rather have a patch :)
See https://github.com/scikit-learn/scikit-learn/blob/master/CONTRIBUTING.md
on how to create one. (The change is tiny, so you don't need to read
the
2014-09-23 6:38 GMT+02:00 c TAKES ctakesli...@gmail.com:
Thanks! What should be the proper behavior when I run the script I wrote?
A MemoryError. The bug is already fixed in master.
--
Meet PCI DSS 3.0 Compliance
2014-09-08 17:53 GMT+02:00 Yaroslav Halchenko s...@onerussian.com:
On Mon, 08 Sep 2014, Yaroslav Halchenko wrote:
hm... actually not clear since it claims that it is because of missing
bdepends
scikit-learn build-depends on missing:
- libsvm-dev (= 2.84.0)
Late reply, but why would it
2014-09-20 21:29 GMT+02:00 c TAKES ctakesli...@gmail.com:
Exception MemoryError: MemoryError() in 'sklearn.tree._tree.Tree._resize'
ignored
Anyone recognize this error?
All too well, but I thought it was fixed for good last time we went
through the tree growing code. Which version exactly are
2014-09-06 10:12 GMT+02:00 Gael Varoquaux gael.varoqu...@normalesup.org:
On Fri, Sep 05, 2014 at 10:12:14PM -0400, Sebastian Raschka wrote:
just saw that scikit-learn 15.2.0 is out and the LDA was fixed. That's great
:).
I missed that reviewing the patches that made it into the 0.15.2.
The
2014-08-28 5:47 GMT+02:00 Josh Wasserstein ribonucle...@gmail.com:
What prior does scikit-learn use for MultinomialNB? The documentation says:
class_prior : array-like, size (n_classes,)
Prior probabilities of the classes. If specified the priors are not adjusted
according to the data.
For
2014-09-04 12:30 GMT+02:00 Lars Buitinck larsm...@gmail.com:
This class prior is just the P(y) in P(y|x) = (P(x|y) × P(y)) / Z.
It's a simple multinomial.
s/multinomial/categorical/ (I always confuse those two
2014-09-04 15:45 GMT+02:00 Karimkhan Pathan karimkhan...@gmail.com:
Oh okay, well I tried with predict_proba. But if query is out of domain then
classifier uniformly divide probability to all learned domains. Like in case
of 4 domains (0.333123570669, 0.333073654046, 0.166936800591,
2014-09-01 17:23 GMT+02:00 Alberto Torres albert...@gmail.com:
I would like to print the coordinate descent coefficients at each iteration.
So far I've identified the code and variable I want to print. In particular
I want to print the variable w in function enet_coordinate_descent from
2014-08-28 11:51 GMT+02:00 Giuseppe Marco Randazzo gmranda...@gmail.com:
I write for you and other people that are interested this tutorial. Hope
will be helpful and more explicative
http://gmrand.blogspot.ch/2014/08/howto-install-scipy-scikit-learn-and.html
Given the frequency with which
2014-08-25 17:12 GMT+02:00 Sheila the angel from.d.pu...@gmail.com:
iris = datasets.load_iris()
gp = {n_neighbors:[2,3], metric:['euclidean']}
clf = GridSearchCV(KNeighborsClassifier(), gp, cv=4).fit(iris.data,
iris.target)
TypeError: __init__() got an unexpected keyword argument 'p'
Why
2014-08-25 17:46 GMT+02:00 Sheila the angel from.d.pu...@gmail.com:
I think the problem is in KNeighborsClassifier not in the grid_search.
You're right. https://github.com/scikit-learn/scikit-learn/pull/3093 fixes it.
2014-08-25 17:50 GMT+02:00 Lars Buitinck larsm...@gmail.com:
2014-08-25 17:46 GMT+02:00 Sheila the angel from.d.pu...@gmail.com:
I think the problem is in KNeighborsClassifier not in the grid_search.
You're right. https://github.com/scikit-learn/scikit-learn/pull/3093 fixes it.
It's just been
2014-08-23 15:44 GMT+02:00 Pavel Soriano sorianopa...@gmail.com:
I don't know if this would be helpful to anybody or if this was already
discussed. That is why I am asking if it is worthy to be pull requested.
Gist URL :
https://gist.github.com/psorianom/0b9d8a742fe0efe0fe82
Yes! BM25 is high
2014-08-23 20:41 GMT+02:00 Gael Varoquaux gael.varoqu...@normalesup.org:
Interesting discussion. Of course, the danger here is that it might be
borderline for the scope of scikit-learn. In case somebody is going to
docstringdo a PR on these topics, I would advise to work on the docstring
and
2014-08-21 12:32 GMT+02:00 Sheila the angel from.d.pu...@gmail.com:
1. What should be the n_jobs value, 8 or (8*4=) 32 ?
n_jobs is the number of CPUs you want to use, not the amount of work.
(It's a misnomer because the number of jobs/work items is variable;
the parameter determines the number
2014-08-21 13:44 GMT+02:00 Joel Nothman joel.noth...@gmail.com:
I think RandomForestClassifier, using multithreading in version 0.15, should
work nested in multiprocessing.
It would work, but the p * n threads from p processes using n threads
each would still compete for the cores, right?
2014-08-19 18:03 GMT+02:00 Adamantios Corais adamantios.cor...@gmail.com:
I am looking for implementations \ configurations of machine learning
algorithms that, instead of a boolean value (class), they return a
probability along with the corresponding confidence error. Any hints?
Any
2014-08-19 20:07 GMT+02:00 Adamantios Corais adamantios.cor...@gmail.com:
Great. And what about the confidence error? I mean, how should I select a
subset of classified data points such that the probability they belong to
any class is high whereas the confidence error is 95% or above?
Sorry, I
2014-08-17 7:26 GMT+02:00 Amey a...@cs.utah.edu:
Zero-one classification Loss as : 0.0426
I would like somebody to help me interpret this in terms of benchmarks :
http://yann.lecun.com/exdb/mnist/
What is the test error % metric given on the link corresponding to the
metrics I have
2014-08-15 2:42 GMT+02:00 Joel Nothman joel.noth...@gmail.com:
This is knowledge you should be able to obtain from almost any machine
learning course or textbook, and you should almost certainly be asking it on
a wider forum than scikit-learn's mailing list, such as
stats.stackexchange.com.
2014-08-12 18:44 GMT+02:00 Saurabh Jha saurabh.j...@gmail.com:
I am trying to fix this issue.
https://github.com/scikit-learn/scikit-learn/issues/3451
Can anyone please recommend some resources for multiclass SVM. I tried to
search for them but got confused. I know these things regarding two
2014-08-06 7:52 GMT+02:00 Joel Nothman joel.noth...@gmail.com:
Instead, could we have an interface in which the `algorithm` parameter could
take any object supporting `fit(X)`, `query(X)` and `query_radius(X)`, such
as an LSHForest instance? Indeed you could also make 'lsh' an available
2014-08-05 14:27 GMT+02:00 Frank Dai soulmach...@gmail.com:
I want to compile scikit-learn under python 3.4, the following commands are
what I'm doing:
alias python=python3
Aliases only work in your current shell, not in make. Use
make PYTHON=python3
2014-08-04 17:39 GMT+02:00 Philipp Singer kill...@gmail.com:
Apart from that, does anyone know a solution of how I can efficiently
calculate the resulting matrix Y = X * X.T? I am currently thinking about
using PyTables with some sort of chunked calculation algorithm.
Unfortunately, this is
2014-07-31 14:04 GMT+02:00 Sheila the angel from.d.pu...@gmail.com:
Also the NearestCentroid classifier do not have decision_function !
I think we should add one, but I've never bothered to figure out what
the right decision function would be. Inverse of distance?
2014-07-29 10:22 GMT+02:00 Eustache DIEMERT eusta...@diemert.fr:
So my question is : is there some utility or snippet to load a CSV into CSR
that I overlooked ?
No, but it's not that hard to write [1].
import array
data = array.array(f)
indices = array.array(i)
indptr = array.array(i,
2014-07-28 23:46 GMT+02:00 Mario Michael Krell kr...@uni-bremen.de:
I have to somehow contradict. In fact it would be possible to get a
probability but it requires some work. So it is not easy.
I my group, we are using a sigmoid fit introduced by Platt to map SVM scores
to probability values.
2014-07-29 14:13 GMT+02:00 Eustache DIEMERT eusta...@diemert.fr:
No, but it's not that hard to write [1].
Do people see any value in either including it in the docs or encapsulating
it into sklearn.datasets ?
If it fits somewhere in the docs, +1 from me. I might offer this to
the SciPy folks
2014-07-29 14:40 GMT+02:00 Joel Nothman joel.noth...@gmail.com:
I think the scipy folks intend that numpy-like setting operations should
suffice for many cases (although be a bit slower than the technique you've
illustrated).
E.g. you can use:
X[i, nonzero] = data[nonzero]
to replace some
2014-07-28 18:39 GMT+02:00 Sheila the angel from.d.pu...@gmail.com:
For the classifier which do not provide probability estimate of the class
(gives error 'object has no attribute predict_proba ), is there any easy
way to calculate the posterior probability?
No. If there were, we would have
2014-07-24 4:35 GMT+02:00 Kartik Kumar Perisetla kartik.p...@gmail.com:
Also, Could someone please throw some light on how HashingVectorizer works?
https://larsmans.github.io/ilps-hashing-trick/
https://en.wikipedia.org/wiki/Feature_hashing
2014-07-23 18:07 GMT+02:00 Michael Eickenberg michael.eickenb...@gmail.com:
To answer 1): yes, if you set cv=number, then it will do K-fold
cross-validation with that number of folds. You can do this explicitly by
using
from sklearn.cross_validation import KFold
cv = KFold(len(data), 6)
2014-07-23 18:21 GMT+02:00 Pagliari, Roberto rpagli...@appcomsci.com:
Is there a way to make prediction, once grid search is done? Right now I’m
getting the error
'GridSearchCV' object has no attribute 'best_estimator_'
Works fine here. What does `python -c 'import sklearn;
2014-07-23 21:31 GMT+02:00 Pagliari, Roberto rpagli...@appcomsci.com:
It says 0.15.0
Right now I am finding the optimal values manually, using cross_validation
(by picking the best average).
That can't be right. This attribute was in place in at least 0.14.0.
How did you install
2014-07-23 0:58 GMT+02:00 Pagliari, Roberto rpagli...@appcomsci.com:
Also, notice that I had to use gs.best_estimator_, and not
gs.best_estimator, and also that the module name for me is sklearn and not
scikits.learn.
Has there been a change in recent versions?
No. Not in recent versions. We
2014-07-16 16:43 GMT+02:00 Andy t3k...@gmail.com:
I'm pretty sure I could use yep for profiling, as mentioned in the docs:
http://scikit-learn.org/dev/developers/performance.html#profiling-compiled-extensions
and get line-by-line counts.
However I did not manage to do that recently. I ususally
2014-07-16 17:29 GMT+02:00 Andy t3k...@gmail.com:
That is using google perftools.
I thought you were referring to the bit about gprof.
So you get line-by-line with google perftools without using debugging
versions? How?
I don't, I look at per-function cost.
2014-07-09 16:11 GMT+02:00 Olivier Grisel olivier.gri...@ensta.org:
Prior to getting this message I notice that the load time of the home
was slow. It might be caused by big thumbnails in the carousel. IIRC,
@larsmans and @jaquesgrobler had worked to reduce the size of the
thumbnails after the
2014-07-09 16:52 GMT+02:00 Lars Buitinck larsm...@gmail.com:
I think that was done on the webserver rather than in version control.
But I just cherry-picked the relevant commit into 0.14.X and pushed
it.
https://github.com/scikit-learn/scikit-learn/pull/3355 has a makefile
target that runs
2014-07-09 19:56 GMT+02:00 Manoj Kumar manojkumarsivaraj...@gmail.com:
Intuitively, a sample with higher weight, should be predicted more
accurately and hence contribute more to the loss. Hence we just multiply,
the sample weight of each term to its loss.
For example, in the least square
2014-07-08 16:00 GMT+02:00 Michael Eickenberg michael.eickenb...@gmail.com:
That totally depends on your data. Here it looks like you are scaling down a
feature that captures a lot of the variation you are looking for, thus
making it less important with respect to the other features in the
2014-07-08 16:27 GMT+02:00 Sheila the angel from.d.pu...@gmail.com:
First I scaled the complete data-set and then splitting it in test and train
data.
Not the cleanest option, but that should work.
--
Open source
2014-07-08 21:33 GMT+02:00 Andreas Mueller t3k...@gmail.com:
On Jul 8, 2014 8:40 PM, Chris Holdgraf choldg...@berkeley.edu wrote:
Hey all - I know that Bootstrap has a billion papers on it, but I was
wondering if there's a specific paper one should reference if we've been
using the Bootstrap
2014-07-04 10:28 GMT+02:00 Olivier Grisel olivier.gri...@ensta.org:
2014-07-04 3:35 GMT+02:00 Dayvid Victor victor.d...@gmail.com:
Should I do the classifier setup in the __init__ (passing all arguments of
the KNN to in the InstanceReduction constructor)?
You might want to pass a KNN instance
2014-07-04 16:37 GMT+02:00 Kyle Kastner kastnerk...@gmail.com:
You should probably read the paper: Training Highly Multiclass Classifiers
http://jmlr.org/papers/v15/gupta14a.html
That said, I think you could gain a lot of value by looking into
hierarchical approaches - training a bunch of
2014-07-03 12:23 GMT+02:00 Kartik Kumar Perisetla kartik.p...@gmail.com:
I am trying to use naive_bayes agorithm for training the model using
partial_fit in scikit-learn.
I tried with 16011( # of features) , 100 training instances and 1018664(
total # of classes), I get an error when I invoke
2014-07-01 21:03 GMT+02:00 Geetu Ambwani geet...@gmail.com:
I imagine this transformer would be useful to others who use lucene for text
analysis and already have access to term vectors and have the partial
pipeline but might still want access to the various weighting schemes
available in
2014-07-01 23:44 GMT+02:00 Joel Nothman joel.noth...@gmail.com:
Calculating TfIdf really isn't that hard. It's much easier for you to do so
while transforming that into DictVectorizer input than for the library to be
everything to everyone.
Indeed. I just indexed 20news in ES, then did
$
2014-07-01 23:58 GMT+02:00 Michael Eickenberg michael.eickenb...@gmail.com:
(the 4th one is typically a kwarg it didn't care about)
Ah:
from elasticsearch import Elasticsearch
es = Elasticsearch()
hits = [es.termvector('20news', 'post', i, fields=['text']) for i in
range(1, 4)]
does the trick,
2014-06-30 0:28 GMT+02:00 Christian Jauvin cjau...@gmail.com:
What explains the difference in terms of the Chi-Square value (0.5 vs 2) and
the P-value (0.48 vs 0.157)?
Here's the feature_extraction.chi2 algorithm:
A = numpy.vstack(([[0,0]] * 18, [[0,1]] * 7, [[1,0]] * 42, [[1,1]] * 33))
X =
2014-06-26 9:15 GMT+02:00 Andy t3k...@gmail.com:
Maybe the calibration is not used for prediction? That would be a bit
odd, though...
That's exactly what's going on. Prediction is consistent with
decision_function, but not predict_proba.
2014-06-26 11:36 GMT+02:00 federico vaggi vaggi.feder...@gmail.com:
Would a pull request clarifying this be welcome, or do people think it's
clear enough as is? Maybe mentioning that this is commonly known as
'loadings' would be enough.
That sounds like a good idea.
2014-06-25 4:50 GMT+02:00 Sturla Molden sturla.mol...@gmail.com:
In general, only POSIX APIs are safe to use on both sides of a fork
Actually, only a short list of async-signal-safe library routines
[1, 2]. Practically all of POSIX is off-limits after fork in a
multithreaded program.
[1]
2014-06-18 19:05 GMT+02:00 Abijith Kp abijith@gmail.com:
I would like to use a Json file from which I take the dataset for my
clustering algorithm. The Json would be in the form of a nested dictionary.
It would be great if someone could show me the correct direction regarding
how to load
2014-06-16 16:56 GMT+02:00 Joel Nothman joel.noth...@gmail.com:
There is, at present, no standard way to do this (although PMML has been
mooted). It depends entirely on which model class you want to export. Which?
Apparently there's a third-party scikit-learn - PMML adapter package now:
2014-06-11 15:54 GMT+02:00 Gavin Gray s0805...@sms.ed.ac.uk:
I need to use Naive Bayes for mixed categorial and numerical data and was
thinking of implementing a flexible Naive Bayes algorithm similar to Weka's
instead of hacking my way around by converting the numerical to categorical
or
2014-06-11 18:16 GMT+02:00 Gavin Gray s0805...@sms.ed.ac.uk:
Yeah, you'd have to hand in a ve ctor listing which distribution to use for
each element in the feature vector. Weka might have a way round this, but
I'll have to try using it to see what the interface is like. They reference
a paper
2014-06-05 13:44 GMT+02:00 ZORAIDA HIDALGO SANCHEZ zora...@tid.es:
Why SelectKBest returns a csr_matrix and is that efficient?
Because it's efficient (compact storage, fast row-wise operations such
as matrix multiplication with CSR on the left), because it's easy to
generate, and because CSR-CSC
2014-06-05 15:04 GMT+02:00 Joel Nothman joel.noth...@gmail.com:
It might also be appropriate to use sklearn.pipeline.FeatureUnion which will
perform an hstack on the output of a number of Transformers. This may be
complicated, but ensures that the transformers are fit correctly,
particularly
2014-06-05 15:12 GMT+02:00 Olivier Grisel olivier.gri...@ensta.org:
Actually I am pretty sure that this will prove problematic as we might
want to have checkins specific for the 0.15.0 release such as version
change, website tweaks and such.
I think it's better to explicitly backport the
2014-06-05 15:24 GMT+02:00 Olivier Grisel olivier.gri...@ensta.org:
I don't think this is possible and this is not what we did for
previous releases:
It's certainly possible, git merge --strategy=ours does a null merge
where the result is just the state of master. The question is whether
we
2014-06-05 15:43 GMT+02:00 Olivier Grisel olivier.gri...@ensta.org:
The question is more in the even we start having release specific
changes to the website such as some done in:
https://github.com/scikit-learn/scikit-learn/commit/9cf67f961e3d0b493f432df0201d91bd0a7dcedf
Right. Ok.
2014-06-02 5:43 GMT+02:00 John Prior john.w.pr...@gmail.com:
Does the optimization really save that much time/space?
Yes, it does. We can now use multithreading instead of separate worker
processes with each a copy of X/y.
There was a PR for a tree-nested dict conversion at some point. I
think
2014-05-30 22:34 GMT+02:00 Anders Aagaard aagaa...@gmail.com:
Which blas implementation are you using? openblas is known to cause this
issue.
Same thought here, but this time it's MKL.
--
Time is money. Stop wasting
2014-05-28 15:57 GMT+02:00 Andrew O'Harney ohar...@gmail.com:
I'm new to the mailing list, so sorry if this is the wrong way to go about
requesting information about work that is underway.
I was wondering if there was any work planned on supporting multiple targets
for the SGDRegression
2014-05-24 0:28 GMT+02:00 Steven Kearnes skear...@gmail.com:
a is a list of the individual DecisionTreeClassifier objects belonging to
the model, instead of a list containing the model itself. The same result
occurs if I add dtype=object to np.asarray.
Why is this happening? Is there a way to
2014-05-22 8:13 GMT+02:00 Gilles Louppe g.lou...@gmail.com:
Just for letting you know, my talk Accelerating Random Forests in
Scikit-Learn was approved for EuroScipy'14. Details can be found at
https://www.euroscipy.org/2014/schedule/presentation/9/.
My slides are far from being ready, but my
2014-05-23 11:08 GMT+02:00 Gilles Louppe g.lou...@gmail.com:
Thanks! Oh, I would be interested in seeing them. Could send me the link if
you still have them?
Here's one with quicksort:
2014-05-23 11:35 GMT+02:00 Gilles Louppe g.lou...@gmail.com:
Thanks! This is really cool! I think I'll try to reproduce some of them and
put one or two in my slides.
I used Fabian's extension_profiler to produce these.
https://github.com/fabianp/extension_profiler
2014-05-21 13:47 GMT+02:00 Olivier Grisel olivier.gri...@ensta.org:
This is a great trick. We might want to get rid of our own partial
copy of CBLAS at some point.
I remember Radim (gensim maint) describing some trouble with BLAS ABIs
on a mailing list some time ago, but I can't find the mail
2014-05-21 13:59 GMT+02:00 Sergio Pascual sergio.pa...@gmail.com:
This is the patch we use in fedora to compile scikit-learn 0.14.1 with
system cblas
http://pkgs.fedoraproject.org/cgit/python-scikit-learn.git/tree/sklearn-unbundle-cblas.patch
But the build system should skip our copy if CBLAS
2014-05-10 19:58 GMT+02:00 Xiandi Zhang zxd_ci...@hotmail.com:
I can run the same successfully in sklearn 0.13. But got this error when I
upgraded to 0.14. This is my first post in this list. If there is anything
I need to specifically provide, in case of reporting an issue, please let me
2014-05-07 9:41 GMT+02:00 Matthieu Brucher matthieu.bruc...@gmail.com:
IMHO GPU will be usable when CPU and GPU memories will be integrated without
move cost. Before, GPU will be hype without mainstream usage.
My thought exactly. Without mapping device memory into the CPU's
virtual memory, no
2014-05-03 11:34 GMT+02:00 Joel Nothman joel.noth...@gmail.com:
... as long as no one ever wrote code like:
for parameters, mean_validation_score, cv_validation_scores in
clf.grid_scores_:
That is actually how tuples are supposed to be used...
I suggest next time we have to return custom
2014-05-01 15:59 GMT+02:00 Ian Ozsvald i...@ianozsvald.com:
Hello. I'm looking at feature_extraction.dict_vectorizer and I'm
wondering why fit() and restrict() use a sorted list of feature names
rather than their naturally-encountered order?
Is there an algorithmic requirement somewhere for
2014-04-28 18:56 GMT+02:00 Jacob Schreiber jmschreibe...@gmail.com:
I understand that HMMs do not perform classification in the same manner as
SVMs or Random Forest, but why is it not desirable to create a new section
to handle HMMs and possibly other graphical models? They seem like an
2014-04-27 18:35 GMT+02:00 Danny Sullivan dsulliv...@hotmail.com:
Is there an intention to add a classifier, like a predict method, to the
BernoulliRBM class?
You mean a discriminative RBM?
--
Start Your Social Network
2014-04-27 19:06 GMT+02:00 Danny Sullivan dsulliv...@hotmail.com:
I see that BernoulliRBM is used primarily as preprocessing to pass off to a
classification algorithm. I initially started thinking about after I saw an
image processing problem using Neural Networks for classification. So my
2014-04-27 18:53 GMT+02:00 Sturla Molden sturla.mol...@gmail.com:
Unlike the NumPy _dotblas module, SciPy uses an f2py wrapper that actually
exports a function poiinter.
Using this scheme to code a fake cblas layer is not difficult either.
I think this question comes about because Manoj is
1 - 100 of 512 matches
Mail list logo