scubamut
10/15/2014 - 4:10 PM

pivots

pivots

pivots.py
content_copyfile_download
# Routine to find pivots of degree k
def pivots(s, k):
    '''Routine to find pivots of degree k in 1D numpy array'''
    peaks = []
    troughs = []
    for i in range(k,len(s) - k):
        #print (s[i], [s[j] for j in range(i-k,i+k+1)])
        if max([s[j] for j in range(i-k,i+k+1)]) == s[i]:
            peaks.append(i)
            #print ('P', i)
        elif min([s[j] for j in range(i-k,i+k+1)]) == s[i]:
            troughs.append(i)
            #print ('T', i)
    
    return peaks, troughs
pivots.md
content_copyfile_download
  • Rendered
  • Source

Definition 1: (Pivot of degree K) The time index t in a timeseries is an upper pivot or a peak of degree K,if for all j =1, 2,...,K, $x_t > x_t−j$ and $x_t > x_i+j$ . Similarly, t is a lower pivot or a trough of degree K,if for j =1, 2,...,K, $x_t < x_t−j$ and $x_t < x_t+j$ .