asNumericMatrix transforms integer columns into numerics ( and marks them as ischar = FALSE ).
matrix2dataFrame only looks for ischar and "factor" and thus does not restore original integer columns back to integer columns ( Those are left as numeric ).
Is this 'not processing integer columns' by 'design' or is this an oversight?
Note, looking at the code, the case seems that boolean columns, also would not be converted to/from boolean: those would also be left as numeric.
asNumericMatrix transforms integer columns into numerics ( and marks them as ischar = FALSE ). matrix2dataFrame only looks for ischar and "factor" and thus does not restore original integer columns back to integer columns ( Those are left as numeric ).
Is this 'not processing integer columns' by 'design' or is this an oversight?
Note, looking at the code, the case seems that boolean columns, also would not be converted to/from boolean: those would also be left as numeric.
Below, see column 'price' (integer).