3 ms·
Sure, if you specify the type. It's still a gotcha because the default behavior is to upcast to floating point unless the type is defined for every integer colu
by jfim 7y ago
Sure, if you specify the type. It's still a gotcha because the default behavior is to upcast to floating point unless the type is defined for every integer column of every data frame, which isn't very pythonic.
The example with the (incorrect) join above shows how even other operations can cause this type conversion.
- qwhelan 7y agoYes, there's a lot of existing code written assuming the old behavior. But most code has only a few ingestion points, so it's pretty simple to turn on.