Examples
Take separate series and convert to numeric, coercing when told to

In [2]:
import numpy as np
import pandas as pd
In [3]:
s = pd.Series(['2.0', '4', -2])
In [4]:
pd.to_numeric(s)
Out[4]:
0    2.0
1    4.0
2   -2.0
dtype: float64
In [5]:
pd.to_numeric(s, downcast='float')
Out[5]:
0    2.0
1    4.0
2   -2.0
dtype: float32
In [6]:
pd.to_numeric(s, downcast='signed')
Out[6]:
0    2
1    4
2   -2
dtype: int8
In [7]:
s = pd.Series(['mango', '2.0', '4', -2])
In [8]:
pd.to_numeric(s, errors='ignore')
Out[8]:
0    mango
1      2.0
2        4
3       -2
dtype: object
In [9]:
pd.to_numeric(s, errors='coerce')
Out[9]:
0    NaN
1    2.0
2    4.0
3   -2.0
dtype: float64