Numpy: Drop rows with all nan or 0 values

Question

I\'d like to drop all values from a table if the rows = nan or 0.

I know there\'s a way to do this using pandas i.e pandas.dropna(how

Answer 1

import numpy as np

a = np.array([
    [1, 0, 0],
    [0, np.nan, 0],
    [0, 0, 0],
    [np.nan, np.nan, np.nan],
    [2, 3, 4]
])

mask = np.all(np.isnan(a) | np.equal(a, 0), axis=1)
a[~mask]

Answer 2

List comprehension can be used as a one liner.

>> a = array([65.36512 , 39.98848 , 28.25152 , 37.39968 , 59.32288 , 40.85184 ,
       71.98208 , 41.7152  , 33.71776 , 38.5504  , 21.34656 , 37.97504 ,
       57.5968  , 30.494656, 80.03776 , 33.94688 , 37.45792 , 27.617664,
       15.59296 , 27.329984, 45.2256  , 61.27872 , 57.8848  , 87.4592  ,
       34.29312 , 85.15776 , 46.37696 , 79.11616 ,       nan,       nan])

>> np.array([i for i in a if np.isnan(i)==False])

array([65.36512 , 39.98848 , 28.25152 , 37.39968 , 59.32288 , 40.85184 ,
       71.98208 , 41.7152  , 33.71776 , 38.5504  , 21.34656 , 37.97504 ,
       57.5968  , 30.494656, 80.03776 , 33.94688 , 37.45792 , 27.617664,
       15.59296 , 27.329984, 45.2256  , 61.27872 , 57.8848  , 87.4592  ,
       34.29312 , 85.15776 , 46.37696 , 79.11616 ])

Answer 3

I like this approach

import numpy as np

arr = np.array([[ np.nan,  np.nan],
                [ -1.,  np.nan],
                [ np.nan,  -2.],
                [ np.nan,  np.nan],
                [ np.nan,   0.]])
mask = (np.nan_to_num(arr) != 0).any(axis=1)

Out:

>>> arr[mask]
... array([[ -1.,  nan],
          [ nan,  -2.]])

Answer 4

This will remove all rows which are all zeros, or all nans:

mask = np.all(np.isnan(arr), axis=1) | np.all(arr == 0, axis=1)
arr = arr[~mask]

And this will remove all rows which are all either zeros or nans:

mask = np.all(np.isnan(arr) | arr == 0, axis=1)
arr = arr[~mask]

Answer 5

In addition: if you want to drop rows if a row has a nan or 0 in any single value

a = np.array([
    [1, 0, 0],
    [1, 2, np.nan],
    [np.nan, np.nan, np.nan],
    [2, 3, 4]
])

mask = np.any(np.isnan(a) | np.equal(a, 0), axis=1)
a[~mask]

Output

array([[ 2.,  3.,  4.]])