I have dataframe with >100 columns. I am trying to select the columns 0~32 and #83. It seems that 1 slice works fine with the code below.
df_new = df[df.columns[0:32]]
It does not work with 2 slices code below though. How do I fix the problem?
df_new = df[df.columns[0:32, 83]]
Use the np.r_ indexer in conjunction with iloc, like this:
df.iloc[:, np.r_[0:32, 83]]
np.r_[0:32, 83]
array([ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16,
17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 83])
df.iloc[:, np.r_[0, -3:]]. If anyone has a solution to that using np.r_ it would be a nice addition to this answer.
c = df.shape[1]; df.iloc[:, np.r_[0, nc-3:nc]]
np.r_), when I extract a column of a dataframe it becomes a DataFrame (type is: pandas.core.frame.DataFrame) instead of a series. Why does this happen? Any way to make this a series?np.r_ is an excellent answer. The other approach would be to construct it with list and range.
Consider this example:
import pandas as pd
df = pd.DataFrame([range(10)],range(10))
cols = list(range(0,5))+[8] # 0,1,2,3,4,8
df.iloc[:,cols]
I am also interest in this problem. We can choose several separate lines or columns. But it seems slicing operation could be done only once on each axis. Like the following one.
new_df=df.iloc[[2,3,4],[3:4]]
Maybe we can slice first and then concatenate them together.
df1=df.iloc[[2:4],:]
df2=df.iloc[[8:10],:]
new_df=pd.concat([df1,df2],axis=0)
