MATLAB Answers

Adam Danz
2

List of built-in data sets, please

Adam Danz
さんによって質問されました 2019 年 11 月 13 日
最新アクティビティ Adam Danz
さんによって コメントされました 2019 年 11 月 25 日 15:21
It would be great if Matlab documentation included a list of built-in data sets. This question has been asked before (2012, 2013, 2014, on statckoverflow) with at least 2200 views in the past 30 days but still I cannot find such a list.
The built-in mat files can be found within the matlabroot (see line of code below) but a list provided in the documentation would be helpful because...
  1. The built-in data sets aren't just in 1 folder; they are distributed across many folders - some folders that we may not think to look in
  2. It's best for users to avoid poking around in the root directories anyway
  3. Not all mat files in the root directories are example data sets so a function that scrapes mat files from root directories isn't efficient
  4. A documented list makes it easy to quickly recall a file name to a data set we are already familiar with
  5. Quickly accessing built-in data makes it easy to test ideas rather than having to generate a fake data set
  6. With a documented list, we can become more familiar with data sets provided by Matlab instead of discovering them after years of daily Matlab use
  7. In the future, we can reference archived documentation to determine when a data set became available
winopen(fullfile(matlabroot,'toolbox','matlab','demos'))
Who do we need to lobby to get this list in the official documentation?

  2 件のコメント

Rik
2019 年 11 月 13 日
Wouldn't it make more sense to post this comment as an answer? That way there is less likelihood of this list being burried under the comments. And is the idea to only add base Matlab data files, or also from different toolboxes?
Adam Danz
2019 年 11 月 13 日
Moved to answers ;)
I started with base Matlab files but toolbox files are good, too. I know the fisheriris data set is often used in the Stats and ML toolbox.

サインイン to comment.

3 件の回答

Steven Lord
回答者: Steven Lord
2019 年 11 月 13 日
 採用された回答

What do you consider a data set?
The census MAT-file in toolbox/matlab/demos (used here among other places) seems like it's obviously a data set. So does outages.csv (used here.)
Are peppers.png and ngc6543a.jpg (used here) also data sets? They're image data, "peppers" appearing on 38 pages in the documentation for MATLAB and 61 times in Image Processing Toolbox. "ngc6543a" occurs less frequently, but still on 19 pages in the documentation for MATLAB.
What about the video file shuttle.avi (used here?)
Does the peaks function (often used in examples to create a simple piece of data to visualize as a surface plot) count as data or code? What about the functions that generate the predefined colormaps or the gallery function?
You can submit an enhancement request to Technical Support asking for such a list to be added to the documentation for one or more products. It would be useful to include in that enhancement request what constitutes a data set in your mind and what information about each data set you'd expect to see in such a list.

  1 件のコメント

Adam Danz
2019 年 11 月 13 日
That's a good point. I realize the 2012 link in my question asks about image data sets (ie, image files such as jpg and png) but what I'm really looking for is a list of mat-file data sets such as the ones listed in the Statistics and Machine Learning Toolbox, Econometrics Toolbox, and Deep Learning Toolbox.. I've used the gallery function and find it very useful but it would be great to have instant access to tables, timetables, cell arrays, strctures, etc. Maybe it's just a problem with my ability to remember the mat file names.
Let's see if there are any bites from the Tech Support.

サインイン to comment.


Adam Danz
回答者: Adam Danz
2019 年 11 月 13 日
編集済み: Adam Danz
2019 年 11 月 15 日

I started a list of built-in mat files and a brief description of their variables (r2019b). Since numeric variables are fairly easy to create on the fly, I only listed files that contain other classes of variables for now. Anyone with edit privileges should feel free to add to this list.
Filename Variable Class size ref Description
-----------------------------------------------------------------------------------
airlineResults.mat result table 29x2 [1] Col1: cell array of airline IDs; Col2: cell array of number of flights per day;
fatalities.mat fatalities table 49x8 [2] Numeric table of highway fatalities and othat statistics by US state
mapredout.mat Key cell 29x1 [3] Cell array of airline initials - some with additional chars
mapredout.mat Value cell 29x1 [3] Cell array of numbers
patients.mat 10-vars (mix) 100x1 [4] Column variables (cells, double, logical) containing information on 100 patients
usastates.mat usastates struct 49x1 [5] Fields: Lon, Lat, Name containing vectors of latitude, longitude, and state name
%% refs
[1] https://www.mathworks.com/help/matlab/import_export/process-big-data-in-matlab-using-mapreduce.html
[2] https://www.mathworks.com/help/matlab/matlab_prog/live-editor-exploratory-programming.html
[3] https://www.mathworks.com/help/matlab/ref/mapreduce.html
[4] https://www.mathworks.com/help/matlab/matlab_prog/split-data-into-groups-and-calculate-statistics.html
[5] https://blogs.mathworks.com/loren/2017/03/28/find-that-signpost-using-optimization-and-the-google-maps-api-in-matlab-to-find-a-landmark/

  0 件のコメント

サインイン to comment.


回答者: Wendy Fullam 2019 年 11 月 25 日 14:59

Just a side note that this request has also been shared with our documentation team for consideration on how to make this easier, going forward.

  1 件のコメント

Adam Danz
2019 年 11 月 25 日 15:21
Thanks, Wendy & the documentation team!

サインイン to comment.



Translated by