table

Methods

⌂ table

Method Reference: table.table2csv

table: table2csv (tbl, file)
table: table2csv (tbl, file, Name, Value)

Write a table to a comma-separated-value (CSV) file.

table2csv (tbl, file) writes the table tbl to file, which may be a character vector, a cellstr, or a string scalar. The resulting file can be read back with csv2table.

The file begins with a comment line reporting how many consecutive rows hold the variable types, names, descriptions, and units, in that order. Those header rows are followed by one row of data per table row.

Variables are serialized as follows:

  • Numeric and logical variables are written as numbers (logicals as 0/1). Missing and infinite values are written as the tokens NaN, NA, Inf, and -Inf.
  • Character, cellstr, and string variables are written as quoted text. A char matrix is one column of text, each row written whole so that its padding survives. A missing string, and every entry of a missing variable, is written as an empty unquoted field, which says that there is no value where "" says the value is an empty string.
  • datetime and duration variables are written in ISO 8601 form, which is exact whatever their display format; the format itself is recorded alongside the variable type, so that it is restored on read. calendarDuration and categorical variables are written as their display strings.
  • A multicolumn variable is split into consecutive columns that share the same variable name.
  • A nested table is split into columns tagged with both the outer and the nested variable name. A structure is split into one column per field, tagged with the variable name and the field name.

When tbl has row names they are written under a leading RowNames column. A row of variable descriptions and a row of variable units are written whenever the corresponding property is set, or a nested variable carries one; a property that was never set writes no row at all, so that it reads back unset.

The following Name-Value options are supported:

NameValue
'WriteVariableNames'A logical scalar specifying whether the variable names are written (default true). When false the file carries none, so csv2table numbers the variables on read and can no longer group the columns: a multicolumn variable comes back as separate variables and a nested table as flat columns.
'WriteRowNames'A logical scalar specifying whether the row labels are written as a leading column (default true).

Source Code: table

Note the following round-trip limitation when reading the file back with csv2table: calendarDuration and categorical variables are returned as cell arrays of character vectors and their values are not reconstructed.

Source Code: table

table2csv writes a table to CSV in the package's round-trippable format: a metadata comment line, then header rows for the variable types, names, and units, followed by one row of data per table row.

 Age = [38; 43; 40];
 Height = [71; 69; 64];
 T = table (Age, Height)
T =
  3x2 table

    Age    Height    
    ___    ______    

     38        71    
     43        69    
     40        64
 T.Properties.VariableUnits = {'Yrs', 'in'};
 filename = fullfile (tempdir (), 'patients.csv');
 table2csv (T, filename);
 type (filename)
"# varTypes 1 rows; varNames 1 rows; varDescriptions 0 rows; varUnits 1 rows.",
"double","double"
"Age","Height"
"Yrs","in"
38,71
43,69
40,64

Because the type and unit metadata is preserved in the header, csv2table reconstructs the table faithfully.

 csv2table (filename)
ans =
  3x2 table

    Age    Height    
    ___    ______    

     38        71    
     43        69    
     40        64
 delete (filename);