如何在Pandas数据框中定义默认缺失值

Question

I want to read a dataframe with given datatype and missing values, but the follwing code is wrong. I have no idea, why this happens!

myText = StringIO("""1,2
3,\N
5,6""")

myDf = pd.read_csv(myText, header=None, names=["a1","a2"], na_values=["\N"], dtype={"a1":"int", "a2":"int"})

I got the error message:

ValueError: Integer column has NA values in column 1

If I remove the dtype option dtype={"a1":"int", "a2":"int"} , then it works fine. Does the integer column don't allow missing values?

Answer 1

Integer doesn't allow missing values. Float allows missing values. If you need it to be integers, you'll need to use a sentinel for the missing ones, like 0 or 99999999 or something (not recommended). Otherwise, use a type like float64 that allows out-of-band values like NaN.