Slicing 1d numpy array with subset of another array

Question

I have an array data which has shape (922, 6) and trgt which is a long 1d time series. data[0] has the start indexes and data[1] the end indexes of subsets that I need to slice from trgt.

I try

trgt[data[:,0:2][0]]
>>> *** IndexError: arrays used as indices must be of integer (or boolean) type

where

data[:,0:2][0]
>>> array([0., 100.])

so I try

trgt[data[:,0:2][0].astype(int)]
>>> array([9909., 9989.])

these are the VALUES at the indexes but not the subset. I try

trgt[tuple(data[:,0:2][0].astype(int))]
>>> *** IndexError: invalid index

how can I get the subset?

DrV · Accepted Answer · 2014-08-20 14:17:29Z

4

This looks odd:

b = data[:, 0:2][0]

because with NumPy arrays it is the same as:

b = data[0, :2]

i.e. the two first columns of the first row of the array. If data is a 2d array, you will get an array with dimensions (2,). It can be used directly to index a 1d vector:

trgt[b]

No need to convert it into a slice or a tuple.

Update: With the updated description of the problem it seems that the first column contains start indices and the second column has the end indices. Then the natural thing to do is to create a list of arrays. 2d-arrays are not a good output format, as then the number of columns should be the same in each row.

In that case a one-liner will do.

lst = [ trgt[row[0]:row[1]] for row in data ]

edited Aug 20, 2014 at 14:17

answered Aug 20, 2014 at 7:32

DrV

23.7k8 gold badges66 silver badges75 bronze badges

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

6 Comments

baconwichsand Over a year ago

data[0, :2] actually has length 992. I am trying to get all 992 subsets. The indexes of the subsets are in data[0] and data[1] respectively

DrV Over a year ago

Are you sure? What does data[0,:2].shape say? Or data.shape?

baconwichsand Over a year ago

sorry I messed up my indexes there, I meant data[:,0:2] -- data[:,0:2].shape = (992,2) and data.shape(992,6)

DrV Over a year ago

If you have a (992,2) array and do data[:,0:2][0], you will get only the first two elements of the first row. If you want to have the first column (a vector of 992 eleemnts), then data[:,0] is what you want. Still, you can index your trgt directly with it.

baconwichsand Over a year ago

I've edited above for better explanation, I am missing something here

|

Collectives™ on Stack Overflow

Slicing 1d numpy array with subset of another array

1 Answer 1

6 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

6 Comments

Your Answer

Sign up or log in

Post as a guest

Related