Rolling or sliding window iterator

Question

I need a rolling window  aka sliding window  iterable over a sequence iterator generator   Default Python iteration can be considered a special case  where the window length is 1   I m currently using the following code   Does anyone have a more Pythonic  less verbose  or more efficient method for doing this   def rolling window seq  window size       it   iter seq      win    it next   for cnt in xrange window size     First window     yield win     for e in it    Subsequent windows         win  -1    win 1           win -1    e         yield win  if   name       main         for w in rolling window xrange 6   3           print w     Example output       0  1  2      1  2  3      2  3  4      3  4  5

User · Answer

def rolling window list  degree       for i in range len list -degree 1           yield  list i o  for o in range degree     Made this for a rolling average function

User · Answer

why not  def pairwise iterable        s - gt   s0 s1    s1 s2    s2  s3            a  b   tee iterable      next b  None      return zip a  b    It is documented in Python doc   You can easily extend it to wider window

User · Answer

There is a library which does exactly what you need   import more itertools list more itertools windowed  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15  n 3  step 3    Out    1  2  3    4  5  6    7  8  9    10  11  12    13  14  15

User · Answer

gt  gt  gt  n  m   6  3  gt  gt  gt  k   n - m 1  gt  gt  gt  print      n   k   format   range i  i m  for i in xrange k     0  1  2   1  2  3   2  3  4   3  4  5

User · Answer

a slightly modified version of the deque window  to make it a true rolling window  So that it starts being populated with just one element  then grows to it s maximum window size  and then shrinks as it s left edge comes near the end   from collections import deque def window seq  n 2       it   iter seq      win   deque  next it  None  for   in xrange 1    maxlen n      yield win     append   win append     for e in it          append e          yield win     for   in xrange len win -1           win popleft           yield win  for wnd in window range 5   n 3       print list wnd     this gives   0   0  1   0  1  2   1  2  3   2  3  4   3  4   4

User · Answer

def GetShiftingWindows thelist  size       return   thelist x x size  for x in range  len thelist  - size   1       gt  gt  a    1  2  3  4  5   gt  gt  GetShiftingWindows a  3     1  2  3    2  3  4    3  4  5

User · Answer

Multiple iterators  def window seq  size  step 1         initialize iterators     iters    iter seq  for i in range size         stagger iterators  without yielding       next iters i   for j in range size  for i in range -1  -j-1  -1       while True           yield  next i  for i in iters            next line does nothing for step   1  skips iterations for step  gt  1           next i  for i in iters for j in range step-1    next it  raises StopIteration when the sequence is finished  and for some cool reason that s beyond me  the yield statement here excepts it and the function returns  ignoring the leftover values that don t form a full window  Anyway  this is the least-lines solution yet whose only requirement is that seq implement either   iter   or   getitem   and doesn t rely on itertools or collections besides  dansalmo s solution

User · Answer

Importing the numpy library import numpy as np arr   np arange 6   Sequence window size   3 np lib stride tricks as strided arr  shape   len arr  - window size  1  window size    strides   arr strides 2      Example output      0  1  2     1  2  3     2  3  4     3  4  5

User · Answer

Trying my part  simple  one liner  pythonic way using islice  But  may not be optimally efficient   from itertools import islice array   range 0  10  window size   4 map lambda i  list islice array  i  i   window size    range 0  len array  - window size   1     output     0  1  2  3    1  2  3  4    2  3  4  5    3  4  5  6    4  5  6  7    5  6  7  8    6  7  8  9     Explanation   Create window by using islice of window size and iterate this operation using map over all array

User · Answer

Optimized Function for sliding window data in Deep learning def SlidingWindow X  window length  stride       indexer   np arange window length  None       stride np arange int len X  stride -window length 4     None      return X take indexer   to apply on multidimensional array import numpy as np def SlidingWindow X  window length  stride1       stride   X shape 1  stride1     window length   window length X shape 1      indexer   np arange window length  None       stride1 np arange int len X  stride1 -window length-1     None      return X take indexer

User · Answer

There s one in an old version of the Python docs with itertools examples   from itertools import islice  def window seq  n 2        Returns a sliding window  of width n  over data from the iterable          s - gt   s0 s1    s n-1     s1 s2     sn                               it   iter seq      result   tuple islice it  n       if len result     n          yield result     for elem in it          result   result 1      elem           yield result   The one from the docs is a little more succinct and uses itertools to greater effect I imagine

User · Answer

How about using the following   mylist    1  2  3  4  5  6  7   def sliding window l  window size 2       if window size  gt  len l           raise ValueError  Window size must be smaller or equal to the number of elements in the list         t          for i in xrange 0  window size           t append l i         return zip  t   print sliding window mylist  3    Output     1  2  3    2  3  4    3  4  5    4  5  6    5  6  7

User · Answer

I tested a few solutions and one I came up with and found the one I came up with to be the fastest so I thought I would share it   import itertools import sys  def windowed l  stride       return zip   itertools islice l  i  sys maxsize  for i in range stride

User · Answer

here is a one liner  I timed it and it s comprable to the performance of the top answer and gets progressively better with larger seq from 20  slower with len seq    20 and 7  slower with len seq    10000  zip   seq i  len seq  - n - 1   i   for i in range n

User · Answer

Modified DiPaolo s answer to allow arbitrary fill and variable step size  import itertools def window seq  n 2 step 1 fill None keep 0        Returns a sliding window  of width n  over data from the iterable          s - gt   s0 s1    s n-1     s1 s2     sn                               it   iter seq      result   tuple itertools islice it  n           if len result     n          yield result     while True                    for elem in it                  elem   tuple  next it  fill  for   in range step           result   result step     elem                 if elem -1  is fill              if keep                  yield result             break         yield result

User · Answer

Here s a generalization that adds support for step  fillvalue parameters   from collections import deque from itertools import islice  def sliding window iterable  size 2  step 1  fillvalue None       if size  lt  0 or step  lt  1          raise ValueError     it   iter iterable      q   deque islice it  size   maxlen size      if not q          return    empty iterable or size    0     q extend fillvalue for   in range size - len q       pad to size     while True          yield iter q     iter   to avoid accidental outside modifications         try              q append next it           except StopIteration    Python 3 5 pep 479 support             return         q extend next it  fillvalue  for   in range step - 1     It yields in chunks size items at a time rolling step positions per iteration padding each chunk with fillvalue if necessary  Example for size 4  step 3  fillvalue         a b c d e f g h i j k l m n o p q r s t u v w x y z   a b c d e f g h i j k l m n o p q r s t u v w x y z   a b c d e f g h i j k l m n o p q r s t u v w x y z   a b c d e f g h i j k l m n o p q r s t u v w x y z   a b c d e f g h i j k l m n o p q r s t u v w x y z   a b c d e f g h i j k l m n o p q r s t u v w x y z   a b c d e f g h i j k l m n o p q r s t u v w x y z   a b c d e f g h i j k l m n o p q r s t u v w x y z   a b c d e f g h i j k l m n o p q r s t u v w x y z        For an example of use case for the step parameter  see Processing a large  txt file in python efficiently

User · Answer

Just a quick contribution   Since the current python docs don t have  window  in the itertool examples   i e   at the bottom of http   docs python org library itertools html   here s an snippet based on the  code for grouper which is one of the examples given   import itertools as it def window iterable  size       shiftedStarts    it islice iterable  s  None  for s in xrange size       return it izip  shiftedStarts    Basically  we create a series of sliced iterators  each with a starting point one spot further forward   Then  we zip these together   Note  this function returns a generator  it is not directly a generator itself    Much like the appending-element and advancing-iterator versions above  the performance  i e   which is best  varies with list size and window size   I like this one because it is a two-liner  it could be a one-liner  but I prefer naming concepts    It turns out that the above code is wrong   It works if the parameter passed to iterable is a sequence but not if it is an iterator   If it is an iterator  the same iterator is shared  but not tee d  among the islice calls and this breaks things badly     Here is some fixed code   import itertools as it def window iterable  size       itrs   it tee iterable  size      shiftedStarts    it islice anItr  s  None  for s  anItr in enumerate itrs       return it izip  shiftedStarts    Also  one more version for the books   Instead of copying an iterator and then advancing copies many times  this version makes pairwise copies of each iterator as we move the starting position forward   Thus  iterator t provides both the  complete  iterator with starting point at t and also the basis for creating iterator t   1   import itertools as it def window4 iterable  size       complete itr  incomplete itr   it tee iterable  2      iters    complete itr      for i in xrange 1  size           incomplete itr next           complete itr  incomplete itr   it tee incomplete itr  2          iters append complete itr      return it izip  iters

User · Answer

This is an old question but for those still interested there is a great implementation of a window slider using generators in this page  by Adrian Rosebrock     It is an implementation for OpenCV however you can easily use it for any other purpose  For the eager ones i ll paste the code here but to understand it better I recommend visiting the original page    def sliding window image  stepSize  windowSize         slide a window across the image     for y in xrange 0  image shape 0   stepSize           for x in xrange 0  image shape 1   stepSize                 yield the current window             yield  x  y  image y y   windowSize 1   x x   windowSize 0      Tip  You can check the  shape of the window when iterating the generator to discard those that do not meet your requirements  Cheers

User · Answer

I use the following code as a simple sliding window that uses generators to drastically increase readability   Its speed has so far been sufficient for use in bioinformatics sequence analysis in my experience   I include it here because I didn t see this method used yet   Again  I make no claims about its compared performance   def slidingWindow sequence winSize step 1      Returns a generator that will iterate through the defined chunks of input sequence  Input sequence must be sliceable            Verify the inputs     if not   type winSize     type 0   and  type step     type 0             raise Exception    ERROR   type winSize  and type step  must be int        if step  gt  winSize          raise Exception    ERROR   step must not be larger than winSize        if winSize  gt  len sequence           raise Exception    ERROR   winSize must not be larger than sequence length           Pre-compute number of chunks to emit     numOfChunks     len sequence -winSize  step  1        Do the work     for i in range 0 numOfChunks step step           yield sequence i i winSize

User · Answer

Just to show how you can combine itertools recipes  I m extending the pairwise recipe as directly as possible back into the window recipe using the consume recipe   def consume iterator  n        Advance the iterator n-steps ahead  If n is none  consume entirely         Use functions that consume iterators at C speed      if n is None            feed the entire iterator into a zero-length deque         collections deque iterator  maxlen 0      else            advance to the empty slice starting at position n         next islice iterator  n  n   None   def window iterable  n 2        s - gt   s0      s n-1     s1      sn    s2       s n 1             iters   tee iterable  n        Could use enumerate islice iters  1  None   1  to avoid consume it  0   but that s       slower for larger window sizes  while saving only small fixed  noop  cost     for i  it in enumerate iters           consume it  i      return zip  iters    The window recipe is the same as for pairwise  it just replaces the single element  consume  on the second tee-ed iterator with progressively increasing consumes on n - 1 iterators  Using consume instead of wrapping each iterator in islice is marginally faster  for sufficiently large iterables  since you only pay the islice wrapping overhead during the consume phase  not during the process of extracting each window-ed value  so it s bounded by n  not the number of items in iterable    Performance-wise  compared to some other solutions  this is pretty good  and better than any of the other solutions I tested as it scales   Tested on Python 3 5 0  Linux x86-64  using ipython  timeit magic   kindall s the deque solution  tweaked for performance correctness by using islice instead of a home-rolled generator expression and testing the resulting length so it doesn t yield results when the iterable is shorter than the window  as well as passing the maxlen of the deque positionally instead of by keyword  makes a surprising difference for smaller inputs     gt  gt  gt   timeit -r5 deque windowkindall range 10   3   0  100000 loops  best of 5  1 87   s per loop  gt  gt  gt   timeit -r5 deque windowkindall range 1000   3   0  10000 loops  best of 5  72 6   s per loop  gt  gt  gt   timeit -r5 deque windowkindall range 1000   30   0  1000 loops  best of 5  71 6   s per loop   Same as previous adapted kindall solution  but with each yield win changed to yield tuple win  so storing results from the generator works without all stored results really being a view of the most recent result  all other reasonable solutions are safe in this scenario   and adding tuple tuple to the function definition to move use of tuple from the B in LEGB to the L    gt  gt  gt   timeit -r5 deque windowkindalltupled range 10   3   0  100000 loops  best of 5  3 05   s per loop  gt  gt  gt   timeit -r5 deque windowkindalltupled range 1000   3   0  10000 loops  best of 5  207   s per loop  gt  gt  gt   timeit -r5 deque windowkindalltupled range 1000   30   0  1000 loops  best of 5  348   s per loop   consume-based solution shown above    gt  gt  gt   timeit -r5 deque windowconsume range 10   3   0  100000 loops  best of 5  3 92   s per loop  gt  gt  gt   timeit -r5 deque windowconsume range 1000   3   0  10000 loops  best of 5  42 8   s per loop  gt  gt  gt   timeit -r5 deque windowconsume range 1000   30   0  1000 loops  best of 5  232   s per loop   Same as consume  but inlining else case of consume to avoid function call and n is None test to reduce runtime  particularly for small inputs where the setup overhead is a meaningful part of the work    gt  gt  gt   timeit -r5 deque windowinlineconsume range 10   3   0  100000 loops  best of 5  3 57   s per loop  gt  gt  gt   timeit -r5 deque windowinlineconsume range 1000   3   0  10000 loops  best of 5  40 9   s per loop  gt  gt  gt   timeit -r5 deque windowinlineconsume range 1000   30   0  1000 loops  best of 5  211   s per loop    Side-note  A variant on pairwise that uses tee with the default argument of 2 repeatedly to make nested tee objects  so any given iterator is only advanced once  not independently consumed an increasing number of times  similar to MrDrFenner s answer is similar to non-inlined consume and slower than the inlined consume on all tests  so I ve omitted it those results for brevity    As you can see  if you don t care about the possibility of the caller needing to store results  my optimized version of kindall s solution wins most of the time  except in the  large iterable  small window size case   where inlined consume wins   it degrades quickly as the iterable size increases  while not degrading at all as the window size increases  every other solution degrades more slowly for iterable size increases  but also degrades for window size increases   It can even be adapted for the  need tuples  case by wrapping in map tuple        which runs ever so slightly slower than putting the tupling in the function  but it s trivial  takes 1-5  longer  and lets you keep the flexibility of running faster when you can tolerate repeatedly returning the same value   If you need safety against returns being stored  inlined consume wins on all but the smallest input sizes  with non-inlined consume being slightly slower but scaling similarly   The deque  amp  tupling based solution wins only for the smallest inputs  due to smaller setup costs  and the gain is small  it degrades badly as the iterable gets longer   For the record  the adapted version of kindall s solution that yields tuples I used was   def windowkindalltupled iterable  n 2  tuple tuple       it   iter iterable      win   deque islice it  n   n      if len win   lt  n          return     append   win append     yield tuple win      for e in it          append e          yield tuple win    Drop the caching of tuple in the function definition line and the use of tuple in each yield to get the faster but less safe version

User · Answer

I like tee     from itertools import tee  izip  def window iterable  size       iters   tee iterable  size      for i in xrange 1  size           for each in iters i                next each  None      return izip  iters   for each in window xrange 6   3       print list each    gives    0  1  2   1  2  3   2  3  4   3  4  5

User · Answer

Let s make it lazy   from itertools import islice  tee  def window iterable  size        iterators   tee iterable  size       iterators    islice iterator  i  None  for i  iterator in enumerate iterators         yield from zip  iterators   list window range 5   3       0  1  2    1  2  3    2  3  4

User · Answer

This seems tailor-made for a collections deque since you essentially have a FIFO  add to one end  remove from the other   However  even if you use a list you shouldn t be slicing twice  instead  you should probably just pop 0  from the list and append   the new item   Here is an optimized deque-based implementation patterned after your original   from collections import deque  def window seq  n 2       it   iter seq      win   deque  next it  None  for   in xrange n    maxlen n      yield win     append   win append     for e in it          append e          yield win   In my tests it handily beats everything else posted here most of the time  though pillmuncher s tee version beats it for large iterables and small windows  On larger windows  the deque pulls ahead again in raw speed   Access to individual items in the deque may be faster or slower than with lists or tuples   Items near the beginning are faster  or items near the end if you use a negative index   I put a sum w  in the body of my loop  this plays to the deque s strength  iterating from one item to the next is fast  so this loop ran a a full 20  faster than the next fastest method  pillmuncher s    When I changed it to individually look up and add items in a window of ten  the tables turned and the tee method was 20  faster  I was able to recover some speed by using negative indexes for the last five terms in the addition  but tee was still a little faster  Overall I would estimate that either one is plenty fast for most uses and if you need a little more performance  profile and pick the one that works best

[python] Rolling or sliding window iterator?

Examples related to python

Examples related to algorithm