Panda .loc or .iloc to select the columns from a dataset

Question

I have been trying to select a particular set of columns from a dataset for all the rows. I tried something like below.

train_features = train_df.loc[,[0,4,5         


        

Answer 1

You can access the column values via the the underlying numpy array

Consider the dataframe df

df = pd.DataFrame(np.random.randint(10, size=(5, 20)))
df

You can slice the underlying array

slc = [0,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18]
df.values[:, slc]

array([[1, 3, 9, 8, 3, 2, 1, 6, 6, 0, 3, 9, 8, 5, 9, 9],
       [8, 0, 2, 3, 7, 8, 9, 2, 7, 2, 1, 3, 2, 5, 4, 9],
       [1, 1, 9, 3, 5, 8, 8, 8, 8, 4, 8, 0, 5, 4, 9, 0],
       [6, 3, 1, 8, 0, 3, 7, 9, 9, 0, 9, 7, 6, 1, 4, 8],
       [3, 2, 3, 3, 9, 8, 3, 8, 3, 4, 1, 6, 4, 1, 6, 4]])

Or you can reconstruct a new dataframe from this slice

pd.DataFrame(df.values[:, slc], df.index, df.columns[slc])

This is not as clean and intuitive as

df.iloc[:, slc]

You could also use slc to slice the df.columns object and pass that to df.loc

df.loc[:, df.columns[slc]]