pandas dataframe index access not raising an exception when out of bounds? - pandas

Question

How can the following MWE script work? I actually want the assignment (right before the print) to fail. Instead it changes nothing and raises no exception. This is some of the weirdest behaviour.
import pandas as pd
import numpy as np
l = ['a', 'b']
d = np.array([[False]*len(l)]*3)
df = pd.DataFrame(columns=l, data=d, index=range(1,4))
df["a"][4] = True
print df

Answer 1

When you say df["a"][4] = True, you are modifying the a series object, and you aren't really modifying the df DataFrame because df's index does not have an entry of 4. I wrote up a snippet of code exhibiting this behavior:
In [90]:
import pandas as pd
import numpy as np
l = ['a', 'b']
d = np.array([[False]*len(l)]*3)
df = pd.DataFrame(columns=l, data=d, index=range(1,4))
df['a'][4] = True
print "DataFrame:"
print df
DataFrame:
a b
1 False False
2 False False
3 False False
In [91]:
df['b'][4]=False
print "DataFrame:"
print df
DataFrame:
a b
1 False False
2 False False
3 False False
In [92]:
print "DF's Index"
print df.index
DF's Index
Int64Index([1, 2, 3], dtype='int64')
In [93]:
print "Series object a:"
print df['a']
Series object a:
1 False
2 False
3 False
4 True
Name: a, dtype: bool
In [94]:
print "Series object b:"
print df['b']
Series object b:
1 False
2 False
3 False
4 False
Name: b, dtype: bool