The objective is to slice multiple pandas row of a specific column using a Numpy boolean.
The following code should do the task
import numpy as np
import numpy.random
import pandas as pd
numpy.random.seed(0)
h=((), ('x'), (), ('y'), (), (), ())
drop_idx=[n for n, dl in enumerate(h) if len(dl)]
df = pd.DataFrame(np.arange(7),columns=['class'])
df.reset_index(inplace=True)
df2=pd.DataFrame(np.arange(5),columns=[('feature','ch1')])
idx_true=np.invert(np.array(h).astype(bool))
g=df[idx_true.tolist()].reset_index(drop=True)
df2['dlabel']=g['class']
However, I wonder whether the above code can be shortened further, especially these lines
idx_true=np.invert(np.array(h).astype(bool))
g=df[idx_true.tolist()].reset_index(drop=True)
df2['dlabel']=g['class']
Currently, Pandas throw an error if I am to directly using Numpy boolean without converting to list
df[idx_true.tolist()]
Is there something I miss, or this is the only way to achieve the intended objective?
>Solution :
You can simply use:
df2['dlabel'] = df.loc[idx_true, 'class'].values
You actually don’t even need to convert h to a numpy array:
df2['dlabel'] = df.loc[[not bool(x) for x in h], 'class'].values
output:
(feature, ch1) dlabel
0 0 0
1 1 2
2 2 4
3 3 5
4 4 6