Converting Pandas Dataframe to numpy array

Question

The following is a Pandas dataframe df

index_column      value_column
 0                   20
 2                   28
 1                   30

It needs to be converted into a numpy array where index_column becomes the index of numpy array and value_column becomes the corresponding value.

That is np_arr[0]=20, np_arr[1]=30, np_arr[2]=28 and so on.

How can this be achieved?

Solution 1

Pandas internally uses np.arrays. The values attribute is all you need:

df.value_column.values

is the np.array that you want.

Solution 2

np_arr = df.value_column.values

Solution 3

np_arr = df.value_column.values

if your column name has special characters like space, use the following:

np_arr = df['temperature [Celsius]'].values

Comments

[ALollz]

And what happens when index_column isn't a RangeIndex starting from 0? numpy array indexing is 0 based, so do you need this to work in general for any arbitrary index arrangement?

[TLanni]

index_column will have 0 to n rows,where n is number of rows in dataframe.But they may not be in ascending order.

[cs95]

1) this has been asked before, and 2) please don't follow any of the answers below if you're using 0.24.

[Ashish kulkarni]

@amandasmith: What you have mentioned is a good way to do it.