Numpy appending to both elements of python list

Question

data = load('data.npy')

def split_classes(data,col):
    newdata = []
    nclasses = len(unique(data[:,col]))
    classes = [[]] * nclasses
    for row in data:
        classes[int(row[col])].append(copy(row))
    print(len(classes[0]),len(classes[1]),len(data))
    return classes
split_classes(data,-1)

This just isn't doing what I want it to do. Values are being added to each list within the python array. The output in this case being: 200 200 200

Example:
Input:
[[4, 2, 0]
 [3, 1, 0]
 [5, 9, 1]]
Output:
[[4, 2, 0],[3, 1, 0]],[5, 9, 1]]

Can you put some example input and output. It's hard to understand what you want to do and what's going wrong without seeing your data before and after... — Adam Hughes
– Adam Hughes, Commented Feb 23, 2015 at 1:20

ely · Accepted Answer · 2015-02-23 01:21:49Z

1

When you use the syntax [[]] * nclasses in Python, it does not mean you get nclasses number of distinct empty list objects. It means you get a list of length nclasses where each element is a handle to the same empty list. If one of them experiences an append operation, they all do.

Instead you could try [[] for i in range(nclasses)]. You can inspect the id of the different elements of classes to verify that indeed they have distinct object ids.

Consider a smaller example:

In [6]: x = [[] for i in range(3)]

In [7]: map(id, x)
Out[7]: [139882701328328, 139882701358288, 139882701358360]

In [8]: x[0].append(1)

In [9]: x
Out[9]: [[1], [], []]

In [10]: y = [[]] * 3

In [11]: y[0].append(1)

In [12]: y
Out[12]: [[1], [1], [1]]

In [13]: map(id, y)
Out[13]: [139882701358216, 139882701358216, 139882701358216]

answered Feb 23, 2015 at 1:21

ely

77.8k36 gold badges158 silver badges234 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

ely Over a year ago

In general, this means that it's OK to use the whole [x] * k trick to get k copies of x whenever x is something immutable, like a string, integer, float, or tuple. But when x is mutable, don't use the [x] * k trick, just explicitly initialize each location in the length-k array.

Collectives™ on Stack Overflow

Numpy appending to both elements of python list

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related