Research Article

A New Fuzzy Logic-Based Similarity Measure Applied to Large Gap Imputation for Uncorrelated Multivariate Time Series

Algorithm 1

FSMUMI algorithm.
Input:: incomplete uncorrelated multivariate time series
: size of time series.
: index of a gap (position of the first missing of the gap)
: size of the gap
: increment for finding a threshold
: increment for finding a similar window
Output:   - completed (imputed) time series
each incomplete signal
each gap at index in
Divide into two separated time series , :
Completing all lines containing missing parameter on by a max trapezoid function
Construct queries , -temporal windows after and before the gap ,
data
Step a: Find the threshold in the database
;
Create a reference window:
(12)Calculate a fuzzy-based similarity measure between and :
(13)Save the to
(14)
(15)end while
(16)return
(17)Step b: Find similar windows in the database
(18);
(19)
(20)
(21)Create a reference window:
(22)Calculate a fuzzy-based similarity measure between and :
(23)if
(24)Save position of to
(25)end if
(26)
(27)end while
(28)return  position of - the most similar window to having the maximum fuzzy similarity measure in the
list.
(29)end for
(30)for data
(31)Perform Step a and Step b for data
(32)return position of - the most similar window to
(33)end for
(34)Replace the missing values at the position by average vector of the window after and the one previous
(35)end for
(36)end for
(37)return - imputed time series