converting series in data frame from exp. object to float-CodePudding

i have csv, and two columns of that csv is exponential notation.

I want to convert it to float or int, and doing this:

df = pd.read_csv('clm_data.csv', sep = ';')
pd.set_option('display.float_format', '{:.0f}'.format)
df.dtypes

those types:

NLS              object
CAMPID           object
RESPONSE_TYPE     int64
RESPONSE_DATE    object
SYSTEM_LK        object

func for transformation:

def unfloater(x):
    value = x.replace(',', '.')
    return float(value)

and when i try to use apply for multiple rows like:

df[["NLS", "RESPONSE_DATE"]] = df[["NLS", "RESPONSE_DATE"]].apply(unfloater)

i got TypeError: cannot convert the series to <class 'float'>

But when i use apply for one series - all works well.

What is wrong with multiple apply? axis = 1 is useless

CodePudding user response：

You are now passing a row to your function instead of a single value, so float() does not work. You can use .astype(float) instead like this:

df = pd.DataFrame({"a": ["1,1e1", "2,2e2", "3,3e3", "4,4e4"],
                   "b": ["5,5e5", "6,6e6", "7,7e7", "8,8e8"]})

df[["a", "b"]].apply(lambda row: row.str.replace(",", ".")).astype(float)

CodePudding user response：

You can use one of the following methods :

Method 1: Use astype()

df['NLS'] = df['NLS'].astype(float)

Method 2: Use to_numeric() to Convert Object to Float

df['NLS'] = pd.to_numeric(df['NLS'], errors='coerce')

to convert all

df = df.apply(pd.to_numeric)
df.dtypes