how to convert str to float in pandas dataframe

Question

I am using below line to read a csv file where column B ends as str format and I do not manage to convert it to float directly:

   df = pd.read_csv('data.csv', sep=";", encoding = "ISO-8859-1")

this produces a dataframe where all columns are in str format:

          A       B
    0   Emma     -20,50
    1   Filo     -15,75
    2   Theo      17,23

As you may notice the decimals are separated by ',' instead of '.' because it's German csv. I tried the following already (to no avail):

  ..., dtype={'B': np.float32}, decimal= ',' , ....

Any idea how I could get it done in the reading process?

Amending after reading the csv is working (but this is an inefficient additional step I would like to avoid), this is what I use:

 df['B'] = df['B'].str.replace(',', '.').astype(float)

Answer 1

For me it works nice, I only omit dtype={'B': np.float32} :

import pandas as pd
import io

temp=u"""A;B
0;Emma;-20,50
1;Filo;-15,75
2;Theo;17,23"""
#after testing replace io.StringIO(temp) to filename
df = pd.read_csv(io.StringIO(temp), sep=";", encoding = "ISO-8859-1", decimal= ',')
print (df)
      A      B
0  Emma -20.50
1  Filo -15.75
2  Theo  17.23

print (df.dtypes)
A     object
B    float64
dtype: object

EDIT:

I think problem can be some decimals are . and some , , then use converters :

import pandas as pd
import io

temp=u"""A;B
0;Emma;-20,50
1;Filo;-15.75
2;Theo;17,23"""


def converter(x):
    return float(x.replace(',','.'))

#define each column
converters={'B': converter}

#after testing replace io.StringIO(temp) to filename
df = pd.read_csv(io.StringIO(temp), 
                 sep=";", 
                 encoding = "ISO-8859-1", 
                 converters=converters)
print (df)

0  Emma -20.50
1  Filo -15.75
2  Theo  17.23

print (df.dtypes)
A     object
B    float64
dtype: object