I would like to import a CSV without specifying the data type of each column before hand.
SemanticImport almost does it, but doesn't appear to recognize numeric columns when there is a mixture of integers and reals. A simple example:
str = "0,0.4\n0,0.3\n0,0\n0.2,0.1"; ds = SemanticImportString[str, HeaderLines -> 0] (* ds[-1,1] = Missing["Unrecognized",0.2] *)
Without specifying the type explicitly, how can I get a number rather than a missing value?