Check if all dataframe row values are in specified range-CodePudding

How to check for each row in dataframe if all its values are in specified range?

import pandas as pd

new = pd.DataFrame({'a': [1,2,3], 'b': [-5,-8,-3], 'c': [20,0,0]})

For instance range <-5, 5>:

>>    a  b   c
>> 0  1 -5  20  # abs(20) > 5, hence no
>> 1  2 -8   0  # abs(-8) > 5, hence no
>> 2  3 -3   0  # abs(-3) <= 5, hence yes

Solution with iteration

print(['no' if any(abs(i) > 5 for i in a) else 'yes' for _, a in new.iterrows()])

>> ['no', 'no', 'yes']

CodePudding user response：

Doing:

out = (df.gt(-5) & df.lt(5)).all(axis=1)
# Or if you just want to supply a single value:
# df.abs().lt(5).all(axis=1)
print(out)

Output:

0    False
1    False
2     True
dtype: bool

You could add this as a new column, and change things to no/yes if desired (which imo is a terrible idea):

df['valid'] = np.where(df.abs().lt(5).all(1), 'yes', 'no')
print(df)

# Output:

   a  b   c valid
0  1 -5  20    no
1  2 -8   0    no
2  3 -3   0   yes

CodePudding user response：

For operations with DataFrames of numbers you should use numpy.

import pandas as pd
import numpy as np


df = pd.DataFrame({'a': [1, 2, 3], 'b': [-5, -8, -3], 'c': [20, 0, 0]})


df_ndarray = df.values

bin_mask = np.where((df_ndarray > 5) | (df_ndarray < -5), 1, 0)

res = np.equal(bin_mask.sum(axis=0), np.arange(len(df.columns)))