People, I updated our Editor Toolbar to reflect the remove whitespace wish of @rhermans
Two small suggestions/request: 1._ When creating a reference link, remove any white space ant the end of the selection. 2._ Allowing the option of using "> " for formatted outputs instead of (* *) comments. Not sure about this one, bu what about also remove Subscript? — rhermans7 hours ago
(I'm not going to implement the second suggestion)
And updated version of the user-script is now online and can be downloaded from the GitHub repo.
Chrome people will be able to update the script from the chrome web store and while you are at it, please rate it!
I received the 10.3 email notice, probably it will come within a week.
"Mathematica 10.3 is scheduled to release soon and your site will be one of the first to get this upgrade. You will be notified by email when the new installers are available for download from the Wolfram User Portal.
Mathematica 10.3 rapidly continues to build out areas first introduced in Version 10, while further refining already best-in-class capabilities. Notable features include new anatomical, linguistic, and geographic data; support to copy, group, and identify entities; increased text processing
@Szabolcs I have a glance at your package, the examples are amazing! Have you considered to make it into an application using WorkBench? In that way, perhaps the examples can be better organized into difference documentation (guide/symbol/tutorial) pages.
@Sosi. I'm not sure what else to say other than your question is statistical in nature best handled in CrossValidated as there doesn't appear to be any issue with how Mathematica performs that test. The Mann-Whitney U test is about attempting to determine if one continuous distribution tends to result in larger (or smaller) values than another (sometimes called statistical dominance).
@xslittlegrass The problem is that I just don't have the time to do that. Ideally of course it would have a proper, integrated documentation. Maybe another time ...
And thank you for testing! Those two errors are normal.
Could someone please check on Windows too? It's Windows I'm worried about. It was painful to compile there.
@JimBaldwin - if I understood correctly, you can only refer to a MW test for comparison of medians if the distributions are similar. My issue raising this in the MMa forum was whether this presumption was tested/validated by the MannWhitneyTest[ ] or not, and if not why there is no warning. If what is being tested are not means, but some other measure of location as you indicate, this should be clearly said in the documentation
@Szabolcs I installed it following the standard installation procedure, and verified that it was on the $UserBaseDirectory. When I first launch it using << IGraphM (tick mark at the end too) I get the following errors:
Get::noopen: Cannot open IGraphM`LTemplate`LTemplatePrivate`. >>
Evaluate IGDocumentation[] to get started.
Import::nffil: File not found during Import. >>
Uncompress::string: String expected at position 1 in Uncompress[$Failed]. >>
Maybe I have not installed it correctly? But I just followed all procedures here (support.wolfram.com/kb/5648)
@Sosi Are you sure you downloaded from this link? It's a zip file. When you extract it, there will be a README.md file next to a directory called IGraphM. This directory should be dropped in $UserBaseDirectory\Applications.
Hmm: ?*`yyyy first had to update from Wolfram servers....What's it doing? Checking the Cloud? Kind of annoying to have to wait for the internet when I've initiated no cloud activity directly. (Who know what M does behind the scenes.)
@Sosi. The way the Mann-Whitney U statistic is constructed, it is all about testing Pr(X<Y) for a single random sample from each distribution. If both distributions have the same shape, then, yes, this is equivalent to testing medians (or any percentile for that matter). But no testing for the same shape is part of the test.
@Sosi. In short, folks (and software) love to ignore assumptions. As a statistician, while I wish it were otherwise, it does keep me in business.
@JimBaldwin or, alternatively, what would you recommend my doing? (btw, the same issue but focused on the statistics side of the question was posted in Cross-validated too: stats.stackexchange.com/questions/176785/…)
After reading the comments here, I felt compelled to install 5.2 on this machine. For some reason, I don't remember the fonts not showing up in the front end the last time I used 5.2, but that was on an older version of Xubuntu, after all.
I have Mathematica 7.01 and Mathematica 5.2 installed on the same machine. I wish to be able to evaluate code in the v.5.2 kernel from within Mathematica 7.01 session. I mean that running Mathematica 7.0.1 standard session I wish to have a command like kernel5Evaluate to evaluate some code in the...
@Sosi. The TTest function performs the Welsh t-test when you ask it to verify the equal variance assumption and then it only seems to do it when that test is rejected. data1 = RandomVariate[NormalDistribution[0, 1], 100]; data2 = RandomVariate[NormalDistribution[0, 2], 100]; t = TTest[{data1, data2}, 0, "HypothesisTestData", VerifyTestAssumptions -> "EqualVariance"] t["DegreesOfFreedom"]. So a "conditional" Welch t-test, yes. "Always perform a Welch t-test", not with TTest.
After WTC, if there's sufficient interest, I'd like to host a 1/2-day or 1-day workshop:
>> WL for data, cloud enablers and emerging business models <<
at this nice Lincon Park, Chicago, Airbnb one of Fri 23--Mon 26 // free pizza, beer, wine courtesy kpg+>statigrafix< // topics >>
Expande...
After WTC, if there's sufficient interest, I'd like to host a 1/2-day or 1-day workshop:
>> WL for data, cloud enablers and emerging business models <<
at this nice Lincon Park, Chicago, Airbnb one of Fri 23--Mon 26 // free pizza, beer, wine courtesy kpg+>statigrafix< // topics >>
Expande...
