How to replace NumPy array values conditionally

Threshold cleanup and labeling tasks often need to change only the values that match a rule. In NumPy, that rule is usually a boolean mask, which keeps each replacement aligned with the array positions instead of moving the work into a Python loop.

Use np.where(condition, x, y) when the replacement should produce a new array. NumPy chooses values from the true branch where the mask is True and from the false branch everywhere else, and each branch can be a scalar or a broadcast-compatible array.

Use boolean mask assignment when an existing array or a deliberate copy should change in place. Copy the source first when later code still needs the original values, and keep compound conditions in a named mask so the threshold and replacement policy remain visible during review.

1. Create a script that compares np.where() with mask assignment on a copy.

   array-replace-conditional.py

   import numpy as np
    
   scores = np.array([35, 72, 88, 41, 93])
    
   low_score = scores < 50
   adjusted = np.where(low_score, 50, scores)
   labels = np.where(scores >= 70, "pass", "review")
    
   in_place = scores.copy()
   in_place[low_score] = 50
    
   print("scores:", scores)
   print("low score mask:", low_score)
   print("np.where adjusted:", adjusted)
   print("mask assignment adjusted:", in_place)
   print("labels:", labels)
   print("changed positions:", np.flatnonzero(adjusted != scores))
   print("source unchanged:", np.array_equal(scores, np.array([35, 72, 88, 41, 93])))
   print("methods match:", np.array_equal(adjusted, in_place))

   np.where() returns adjusted as a new array. The mask assignment line writes into in_place because that variable is an explicit copy of scores.

2. Run the script.

   $ python array-replace-conditional.py
   scores: [35 72 88 41 93]
   low score mask: [ True False False  True False]
   np.where adjusted: [50 72 88 50 93]
   mask assignment adjusted: [50 72 88 50 93]
   labels: ['review' 'pass' 'pass' 'review' 'pass']
   changed positions: [0 3]
   source unchanged: True
   methods match: True

3. Verify that both replacement methods changed the same positions.

   changed positions: [0 3] identifies the scores below 50. methods match: True confirms np.where() and mask assignment produced the same adjusted values.

4. Confirm that the source array remained unchanged.

   source unchanged: True means the original scores array still contains [35 72 88 41 93]. Use mask assignment only on an array that should be mutated.