How to get sample datasets locally? [Documentation error?]

Now that documentation is offline by default, it seems that the sample datasets mentioned on this documentation page are no longer available locally, either. As a result, those files can no longer be accessed by the method specified on that page. For example,

load fisheriris.mat

will not find that file.

Is there a way to access those sample datasets without downloading the entire documentation? I hope so!

(At the very least, that documentation page should be updated.)

[Ive J]

it still works on my R2023a [academic licence] 🤔

data = load("fisheriris.mat")

data = struct with fields:
    species: {150×1 cell}
       meas: [150×4 double]

[the cyclist]

Your institution presumably has the full documentation locally (which is no longer the default configuration).

Answer 1

[Cris LaPierre]

Beginning in R2023a, the product documentation is no longer installed by default (see here). The side effect of this is that some example datasets are no longer included with the product install.

This means some example datasets are no longer available by default. You can still access them by either installing the documentation, or by finding an example in the online documentation that uses that file, and then use the 'Open in MATLAB Online' or 'Copy Command' workflows to download that example (along with any associated data files) to a local folder and open it in your MATLAB instance.

Here are 3 examples that use fisheriris.mat

• Specify Variable-Size Arguments for Code Generation
• Interpret Deep Network Predictions on Tabular Data Using LIME
• Code Generation for Nearest Neighbor Searcher

The copy command from one of them is openExample('stats/SpecifyVariableSizedArgumentsForCodeGenerationExample')

Here's a screen shot of what it looks like loading an example that uses the ovariancancer.mat dataset.

If it is a file you use often, consider adding the example folder to your MATLAB Path.

[the cyclist]

@Cris LaPierre, I appreciate your detailed answer. I have it working.

My question may have been a red herring. I was inadvertently using the 2023b Prerelease version, which I think may have been the problem. Using the openExample() method did not work in that version, but it did in 2023a. It seems that the stats sample datasets may even be present in 2023a (obviating the need for all of the above), but now I'm not sure. I don't feel like re-installing 2023a to find out, but when I download 2023b (the real one, not the prerelease), I'll see how everything works, and revisit my question here.

Thanks for your help!

[Cris LaPierre]

I also have tested in the prerelease (update 2), and the dataset is present, so no need to use the openExample workflow.

>> which fisheriris.mat
C:\Program Files\MATLAB\R2023b\toolbox\stats\statsdata\fisheriris.mat

[the cyclist]

Interesting. I'm running Update 1, for macOS (Apple Silicon). I don't see that I have access to Update 2 yet.