4

If I have a pandas.DataFrame in python I can inspect the data types for the DataFrame with the dtypes attribute. How can I do the same with a Matlab table? I poked around the Properties mechanism but didn't find anything type oriented there.

jxramos
  • 7,356
  • 6
  • 57
  • 105
  • Isn't there a Property called `VariableTypes`? https://www.mathworks.com/help/matlab/ref/table.html – rahlf23 Oct 01 '18 at 20:12
  • If I try a `.` tab completion after Properties I get presented with `Description`, `DimensionNames`, `RowNames`, `UserData`, `VariableContinuity`, `VariableDescriptions`, `VariableNames`, and `VariableUnits`. I'm in Matlab 2018a. Upon reading the doc it seems VariableTypes is an input string to coerce a data type. – jxramos Oct 01 '18 at 20:16
  • Looks like there's also a [vartype](https://www.mathworks.com/help/matlab/ref/vartype.html) function which grabs variables from a `table` of a certain type. But that's a different operation too. I'm trying to do something more introspective than something selective. – jxramos Oct 01 '18 at 20:22

2 Answers2

3

It appears that you can call the following:

varfun(@class,t,'OutputFormat',table)


class_var1   class_var2    class_var3    class_var4 
__________   __________    __________    __________

double       double        double        double

where t is your table. I am referencing the answer here.

Further documentation on varfun is available here as well.

jxramos
  • 7,356
  • 6
  • 57
  • 105
rahlf23
  • 8,869
  • 4
  • 24
  • 54
  • Nice find, its choking on my particular table, but I think because it has some nested matrix type or something. When I inspect it in the DataInspector there are some "merged" columns for some of the variables, eg variable 2 spans 4 columns: `class(t) ; ans = 'table'; varfun(@class,t,'OutputFormat',cell) ; Error using cell Not enough input arguments.` – jxramos Oct 01 '18 at 20:28
  • 2
    @jxramos: If you want to find the class for an individual column, you can also simply say `class(t.var1)`. – Cris Luengo Oct 01 '18 at 20:59
0

Looks like one way to get this information (while unfortunately getting a bunch of other things) is with a simple call to summary.

Here's some sample output

K>> summary(t)

Variables:

    var1: 2966185×1 double

        Units:  sec
        Values:

            Min       56.207
            Median    7466.7
            Max        14878

    var2: 2966185×4 double

        Values:
                        var2_1        var2_2      var2_3      var2_4
                      ________    __________    ________    ________

            Min       -0.99966      -0.99901    -0.99887    -0.99998
            Median     0.01644    -0.0044018     0.12838      0.1564
            Max        0.98176       0.96433     0.99998           1

    var3: 2966185×3 double

        Units:  g
        Values:
                         var3_1        var3_2        var3_3
                      __________    _________    __________

            Min           -2.779      -3.1366       -3.6089
            Median    -0.0002124    -0.002221    -0.0020435
            Max           3.7874       5.9634        2.8443


    var4: 2966185×1 double

        Values:

            Min          0
            Median       5
            Max          5
jxramos
  • 7,356
  • 6
  • 57
  • 105
  • Is this telling the data type though? – rahlf23 Oct 01 '18 at 20:32
  • @rahlf23, sure, eg `var1: 2966185×1 double`, it's a double column vector, and var 2 and 3 are matrices of double. That appears to be what broke the `varfun` solution which probably handles single dimension variables. – jxramos Oct 01 '18 at 20:34