import numpy as np
import pandas as pd
from IDK_T import IDK_T
from IDK_square_sliding import IDK_square_sliding


# Read data. 
# The period length (cycle) of each time series and the locations of anomalous period subsequences are given in the appendix of the paper.
X=np.array(pd.read_csv("Discords_Data/noisy_sine.txt",header=None)).reshape(-1,1)
cycle=300


# IDK square using non-overlapping windows. The input series should have single or multiple periodicity.
# The period length (cycle) of the time series needs to be input. 
# The number of partitionings t is 100 by default. The sample sizes psi1 and psi2 for two levels of IK mappings need to be searched
# and the search range we use is provided in the paper.
# Return a list of similarity scores for each period of time series. Anomalous period subsequences are those having the lowest similarity scores.
similarity_score=IDK_T(X,t=100,psi1=16,width=cycle,psi2=2)


# IDK square using a sliding window. The input series can be periodic or aperiodic time series showing recurring normal subsequences.
# The length of the sliding window needs to be searched and the search range we use is provided in the paper.
# Return a list of similarity score for each subsequence of time series extracted by the sliding window. Anomalous subsequences are those having the lowest similaity scores.
w=cycle-50
sliding_score=IDK_square_sliding(X,t=100,psi1=4,width=w,psi2=4)