CODEWITHSUNDEEP
pythontuplesnan
zelda26
zelda26

Reputation: 489

python tuple remove nan

I have a list that contains tuples. I need to remove any tuple that contains nan. How can I do this?

The type of nan is math.isnan. It returns True.

math.isnan(tmp[1][1][1])

I tried this and it does not work:

import math
res = [t for t in tmp if not any(isinstance(n, float) and float('nan') for n in t)]
print(res)

I also tried this and it does not work:

nans = {np.nan, math.nan, float('nan')}
[x for x in tmp if not nans.intersection(x)]

here is an example tuple

tmp = [[(0, 1.0),
  (5, nan),
  (10, 1.0),
  (15, nan),
  (20, 1.0),
  (25, nan),
  (30, 1.0),
  (35, nan),
  (40, 0.9808612440191388),
  (45, nan),
  (50, 0.9719298245614035),
  (55, nan),
  (60, 0.9712918660287081),
  (65, nan),
  (70, 0.8947368421052632),
  (75, nan),
  (80, 0.956140350877193),
  (85, nan),
  (90, 0.6140350877192983)],
 [(0, 0.9635627530364372),
  (5, nan),
  (10, 1.0),
  (15, nan),
  (20, 0.9924812030075187),
  (25, nan),
  (30, 1.0),
  (35, nan),
  (40, 0.9808612440191388),
  (45, nan),
  (50, 0.9508771929824561),
  (55, nan),
  (60, 0.9617224880382775),
  (65, nan),
  (70, 0.8173374613003096),
  (75, nan),
  (80, 0.8464912280701754),
  (85, nan),
  (90, 0.5614035087719298)]]

Upvotes: 0

Views: 1346

Answers (1)

James
James

Reputation: 36691

You can do this with a nested list comprehension. Iterate over each sublist of the list, iterate over each tuple, check if the tuple contains a nan.

[[tup for tup in sublist if not any(map(math.isnan, tup))] for sublist in tmp]
# returns:
[[(0, 1.0),
  (10, 1.0),
  (20, 1.0),
  (30, 1.0),
  (40, 0.9808612440191388),
  (50, 0.9719298245614035),
  (60, 0.9712918660287081),
  (70, 0.8947368421052632),
  (80, 0.956140350877193),
  (90, 0.6140350877192983)],
 [(0, 0.9635627530364372),
  (10, 1.0),
  (20, 0.9924812030075187),
  (30, 1.0),
  (40, 0.9808612440191388),
  (50, 0.9508771929824561),
  (60, 0.9617224880382775),
  (70, 0.8173374613003096),
  (80, 0.8464912280701754),
  (90, 0.5614035087719298)]]

Upvotes: 4

Related Questions