28,130 reputation
259135
bio website
location Argentina
age
visits member for 2 years, 10 months
seen 39 mins ago

My email:

StringTake[ ExampleData[{"Text", "Hamlet"}] <> "@", List /@ {45, 114, 132, 134, 138, 142, 2933, 2939, -1, 2994, 3184, 3189, 3213, 3305, 3364, 3427, 3446, 3473}] // StringJoin


Aug
8
comment ClassifierInformation's “ClassPriors” misreporting?
Great, thank you
Aug
7
comment ClassifierInformation's “ClassPriors” misreporting?
@MichaelE2 I just intended to mean that that section was for those that would say too-long-didn't-read to a long post, and the second to those that would complain of a lazy incomplete question.
Aug
7
revised ClassifierInformation's “ClassPriors” misreporting?
added 2 characters in body
Aug
7
answered ClassifierInformation's “ClassPriors” misreporting?
Aug
7
revised ClassifierInformation's “ClassPriors” misreporting?
edited tags
Aug
7
comment Accessing list elements by name
And maybe change Association@Thread to AssociationThread
Aug
7
comment ClassifierInformation's “ClassPriors” misreporting?
@rcollyer great, thanks!
Aug
7
comment V10's Operator Forms - what are they good for?
The ~(#2 /@ #1 &)~ form is nicer than the //Max/@#& in that you keep the ability to continue with other infixes like ~Position~2 without having precedence problems. +1
Aug
7
comment Classify the classic tennis example
Overfitting so that the original points all match is something undesirable, but it depends on the method. It shouldn't happen with nearest neighbor (with a single neighbor)
Aug
7
comment ClassifierInformation's “ClassPriors” misreporting?
Also, these machine learning functions seem to be very picky about receiveng the options/parameters as strings or as symbols. I most of mma, the string versions work even if the symbol exists, but not here. To make matters worse, ClassPriors is a symbol as an option and a string as a property (but not all properties are strings, such as Method). I'll probably report this since it is pointlessly annoying and inconsistent with other mma functions
Aug
7
asked ClassifierInformation's “ClassPriors” misreporting?
Aug
7
comment Using MapIndexed[List] on a dataset fails — why?
Similar to Dataset[{True}][Apply[And]]. I wonder if we are missing something
Aug
6
comment Combining images with CUDA
I wonder if cuda would be faster than Simon's suggestion given that most of the time must be spent copying to GPU memory, I doubt it. Ahh, I need to get cuda working on linux
Aug
6
revised Convert a list of rules into 2 lists of integers
added 36 characters in body
Aug
6
answered Convert a list of rules into 2 lists of integers
Aug
6
comment Using MapIndexed[List] on a dataset fails — why?
This looks like a bug to me
Aug
5
comment How to merge Datasets representing matrices with named rows/columns?
@TaliesinBeynon definately. Combining all, merge[level_] := Nest[Merge, Identity, level]@*Map[Normal]@*List /* Dataset. And merge[2][ds1, ds2]
Aug
5
comment How to merge Datasets representing matrices with named rows/columns?
Is this simple enough? Merge[Normal/@{ds1, ds2}, Merge[#, Identity] &] // Dataset
Aug
5
comment List of atomic expressions
Dataset is atomic
Aug
5
comment Are there any tutorials on version 10 notebook templating and report generation? Are there missing docs?
Glat to hear it's coming