从Python中的另一个数组切片给定范围的数组

Question

To be more clear, let me reword my question that I initially asked below.

I have a series of data points that rise and fall in amplitude as someone takes a step and then another step. Zero is when the foot is off the ground.

A simple example would look like this:

 data_array = (0,0,0,10,20,50,40,30,10,0,0,0,0,0,10,20,50,40,30,10,0,0)

I determine when each step starts and stop by recording the starts (all the indexes that start) in one array and the stops in another array.

starts = (4 15)

stops = (9 20)

THE QUESTION: Now, I want to slice the actual data for each step out of the initial array and them in columns.

[Note: That, if necessary, we know the number of steps taken by the amount of data in the starts or stops arrays.]

10 10

20 20

50 50

40 40

30 30

10 10

I can not figure out how to use those starts and stops indices to slice the initial array. OR, I did not find a filter function to slice out the steps.

BTW (2nd Edit) Here is some of the code I am using if it helps:

sigma = 5

threshold = 30

above_threshold = gaussian_filter(Fz, sigma=sigma) > threshold

#---INDEX ALL STATE CHANGES---

ind = np.where(np.diff(above_threshold))[0] + 1

print ind

ORIGINAL QUESTION

In this type of array:

data_array = (0,0,0,10,20,50,40,30,10,0,0)

I determine when values are above 20 and below 20. I return those indexes as starts (going above 20) and stops (going below 20) by making the following:

startstop = np.vstack((ind[::2], ind[1::2])).T   
starts1=np.vstack((ind[::2])).T
stops1=np.vstack((ind[1::2])).T

Can someone point me in the right direction using numpy (or not) so that I can extract all those values in data_array using one of these arrays (startstop, starts1, stops1) to get this:

new_array = (50,40,30)

Thanks, Scott

Answer 1

I don't understand the way you are trying to go about it, but can you use filter ?:

>>> data_array = (0,0,0,10,20,50,40,30,10,0)
>>> filter(lambda x: x>20,data_array)

(50, 40, 30)

Works with numpy arrays too.

Answer 2

take() will do what you wants, i think.

data_array = np.array([0,0,0,10,20,50,40,30,10,0])


b = data_array.take([1,2,3])
print b

output:

[ 0  0 10]

read more: take() more on take

Answer 3

Your question is not perfectly clear, but if you are asking how to slice a Python list or iterable, use the built-in slice function :

>>> data_array = (0,0,0,10,20,50,40,30,10,0)
>>> data_array[slice(5,8)]
(50, 40, 30)
>>> new_list=list(data_array[slice(5,8)])
>>> new_list
[50, 40, 30]

Answer 4

Python lists are defined with square brackets, and we want to generate a list of lists (where each piece contains one of your defined segments). Since computers start counting at 0, your "4th element is the start" translates to array index = 3.

One quirk is that to query the 4th to 9th element we'd use data_array[3:9]: this slice notation gives you every element starting at the first specified, up to (but not including) the last. The rest is a list comprehension that goes over any arbitrary number of step segments.

starts = [4, 15]
stops = [9, 20]
data_array = [0,0,0,10,20,50,40,30,10,0,0,0,0,0,10,20,50,40,30,10,0,0]
segments = [ data_array[starts[i] - 1: stops[i]  ]  for i in range( len(starts) ) ]

returns

>>> segments
[[10, 20, 50, 40, 30, 10], [10, 20, 50, 40, 30, 10]]

Each set of steps (segment) can be accessed individually:

>>>segments[0]
[10, 20, 50, 40, 30, 10]

EDIT: Alternately... if you need to use numpy array, then try: segments = array([ data_array[starts[i] - 1: stops[i] ] for i in range( len(starts) ) ])

I tried working in additional modules to get the array directly (using itertools.islice with itertools.chain , numpy.fromiter , etc). But even if the fancier solution worked, I'm not sure it'd offer significant speed advantages over converting to an array, and it would be a lot less concise. See: How do I build a numpy array from a generator?