Round all values in table

Hi,

I have a table with calculated values. Now I want to round all values to 2 decimals. How can I do that in once for the whole table?

Answer 1

[Scott MacKenzie]

I think this achieves what you are after:

% test data
T1 = array2table(rand(5))

T1 = 5×5 table
      Var1        Var2       Var3        Var4        Var5  
    _________    _______    _______    ________    ________

    0.0024079    0.97172    0.41667    0.027209    0.079337
      0.14919    0.17995    0.58257     0.46413     0.99968
      0.56165    0.79083    0.06821     0.18682     0.23824
      0.31412     0.6835    0.82666     0.49151     0.12438
      0.78186    0.66873    0.07059     0.86243     0.32011

% rounded to 2 decimal places
T2 = array2table(round(T1{:,:},2))

T2 = 5×5 table
    Var1    Var2    Var3    Var4    Var5
    ____    ____    ____    ____    ____

       0    0.97    0.42    0.03    0.08
    0.15    0.18    0.58    0.46       1
    0.56    0.79    0.07    0.19    0.24
    0.31    0.68    0.83    0.49    0.12
    0.78    0.67    0.07    0.86    0.32

Answer 2

[Andrew Sandeman]

Another solution using varfun()

% test data
T = array2table(rand(5))

T = 5×5 table
       Var1         Var2         Var3         Var4        Var5  
    __________    _________    _________    ________    ________

       0.10584      0.27303    0.0067048     0.58837    0.063311
       0.45211      0.12874      0.22154      0.4381     0.57343
       0.59635      0.64793      0.89923     0.64467     0.48561
       0.75997    0.0090159      0.33324     0.42146     0.60051
    0.00011342      0.11944      0.35647    0.059849    0.066758

% round data in table to 2 d.p.
varfun(@(x)round(x,2), T)

ans = 5×5 table
    Fun_Var1    Fun_Var2    Fun_Var3    Fun_Var4    Fun_Var5
    ________    ________    ________    ________    ________

      0.11        0.27        0.01        0.59        0.06  
      0.45        0.13        0.22        0.44        0.57  
       0.6        0.65         0.9        0.64        0.49  
      0.76        0.01        0.33        0.42         0.6  
         0        0.12        0.36        0.06        0.07  

The only difference is the variable names.

[Walter Roberson]

If you use the varfun() option InputVariables=@numeric then you can have the rounding apply only to the numeric portions of the table.

However, variables not selected by the @isnumeric will simply not appear in the output table, and the variables that do appear will have 'round_' as a prefix. This is not exactly the most convenient for post-processing.

Another approach is to create a helper function along the lines of

function out = conditional_round2(in)
   if isnumeric(in)
       out = round(in,2);
   else
       out = in;
   end
end

and then

new_T = varfun(@conditional_round2, T);
new_T.Properties.VariableNames = regexprep(T.Properties.VariableNames, '^[^_]*_', '', 'once');

[Andrew Sandeman]

Thanks very much, I had no idea about the InputVariables option to varfun(), something I definitely could have used in the past!

What about this, still using varfun():

% test data
t = table(rand(5,1), repelem("a",5,1))

t = 5×2 table
      Var1      Var2
    ________    ____

      0.4024    "a" 
     0.21798    "a" 
    0.070511    "a" 
      0.1205    "a" 
     0.72754    "a" 

% round data in table to 2 d.p.
t(:,"Var1") = varfun(@(x)round(x,2), t, InputVariables=@isnumeric)

t = 5×2 table
    Var1    Var2
    ____    ____

     0.4    "a" 
    0.22    "a" 
    0.07    "a" 
    0.12    "a" 
    0.73    "a" 

The other method, whilst also good, for me coming up with that regex pattern is still a mental speedbump. However, I can see an extension to conditionalRound. I would call it a conditionalFunction, and should you want to say round to 2 d.p. if is column is numeric, do something else if isstring e.g.

function out = conditionalFunction(in)
   if isnumeric(in)
       out = round(in,2);
   elseif isstring(in)
       % do something to strings
   else
       out = in;
   end
end

[Walter Roberson]

That's an interesting use of table assignment.

Needs more work if there could be multiple output variables.

t = table(rand(5,1), rand(5,1), repelem("a",5,1))

t = 5×3 table
     Var1       Var2      Var3
    _______    _______    ____

    0.89086    0.66185    "a" 
    0.83846    0.33959    "a" 
    0.78412    0.42468    "a" 
    0.34779    0.68343    "a" 
    0.49041    0.28647    "a" 

outvars = t.Properties.VariableNames(varfun(@isnumeric,t,OutputFormat="uniform"))

outvars = 1×2 cell array
    {'Var1'}    {'Var2'}

t(:,outvars) = varfun(@(x)round(x,2), t, InputVariables=outvars)

t = 5×3 table
    Var1    Var2    Var3
    ____    ____    ____

    0.89    0.66    "a" 
    0.84    0.34    "a" 
    0.78    0.42    "a" 
    0.35    0.68    "a" 
    0.49    0.29    "a" 

[Andrew Sandeman]

And here it is wrapped up in a function.

Awesome! I feel like with some documentation it might be worth putting on the file exchange?

t = table(rand(5,1), rand(5,1), repelem("a",5,1))

t = 5×3 table
     Var1        Var2      Var3
    _______    ________    ____

    0.45139     0.33976    "a" 
     0.6767     0.59527    "a" 
    0.46407    0.070445    "a" 
    0.87703     0.88928    "a" 
    0.30553     0.68711    "a" 

conditionalVarfun(t, @isnumeric, @(x)round(x,2))

ans = 5×3 table
    Var1    Var2    Var3
    ____    ____    ____

    0.45    0.34    "a" 
    0.68     0.6    "a" 
    0.46    0.07    "a" 
    0.88    0.89    "a" 
    0.31    0.69    "a" 

%% functions

function t = conditionalVarfun(t, fh_condition, fh_operation)
% t (table)
% fh_condition e.g. @isnumeric
% fh_operation e.g. @(x)round(x,2)

p = inputParser;
p.addRequired("t", @istable)
p.addRequired("fh_condition", @isfunctionhandle)
p.addRequired("fh_operation", @isfunctionhandle)
p.parse(t, fh_condition, fh_operation)
t = p.Results.("t");
fh_condition = p.Results.("fh_condition");
fh_operation = p.Results.("fh_operation");

outvars = t.Properties.VariableNames(varfun(fh_condition,t,OutputFormat="uniform"));
t(:,outvars) = varfun(fh_operation, t, InputVariables=outvars);

end

function bool = isfunctionhandle(x)
bool = isa(x, 'function_handle');
end