STMotif R Package

Heraldo Borges, Amin Bazaz, Esther Pacitti, Eduardo Ogasawara

2020-11-14

The goal of the STSMotif R package is to allows the discovery and ranking of a motif in spatial-time series quickly and efficiently.

Introduction

A pattern that significantly occurs in a time series is called a motif. In spatial time series data, these patterns may not be substantially present in a single time series but dispersed over several times series, limited in both space and time. The STMotif R package was developed to simplify the Spatio-temporal data mining on the search for these motifs. We present the functions available in STMotif package through the sample dataset, also available in this package.

First, install the package by typing:

install.packages("STMotif")

Then, load the package by typing:

library(STMotif)

It provides two categories of functions: for discovering and ranking motifs (CSAMiningProcess) and functions for viewing the identified motifs.

1. CSAMiningProcess

  1. The function NormSAX allows the normalization and SAX indexing of the dataset.

# The process is launched on the provided example dataset
dim(D <- STMotif::example_dataset)
#> [1] 20 12

# Normalizartion and SAX indexing
DS <- NormSAX(D = STMotif::example_dataset,a =5)

# Information of the normalized and SAX indexing dataset 
# The candidates built 
head(NormSAX(D = STMotif::example_dataset, a = 5)[,1:10])
#>                      
#> 1 a c c c c c c c e c
#> 2 a a e c e e e c c e
#> 3 c e e e c e d e e e
#> 4 e e b e e d e e d b
#> 5 e c c b b c b c a e
#> 6 b d c a a a b e a d
  1. The function SearchSTMotifs allows to check and filter the stmotifs, grouping the motifs from the neighboring block.
# The list of motifs 
# stmotifs <- SearchSTMotifs(D,DS,w,a,sb,tb,si,ka)
stmotifs <- SearchSTMotifs(D,DS,4,5,4,10,2,2)
stmotifs[[1]]
#> $isaxcod
#> [1] "ceeb"
#> 
#> $recmatrix
#>      [,1] [,2] [,3]
#> [1,]    1    0    0
#> [2,]    0    0    0
#> 
#> $vecst
#>   s t
#> 1 1 3
#> 2 3 1
#> 3 4 2
  1. The function RankSTMotifs allows to rank the stmotifs list, making a balance between distance among the occurrences of a motif with the encoded information on the motif itself and his quantity.
# The rank list of stmotifs 
rstmotifs <- RankSTMotifs(stmotifs)
rstmotifs[[1]]
#> $isaxcod
#> [1] "bded"
#> 
#> $recmatrix
#>      [,1] [,2] [,3]
#> [1,]    0    0    0
#> [2,]    1    1    1
#> 
#> $vecst
#>    s  t
#> 1  1 11
#> 2  2 11
#> 3  4 17
#> 4  5 17
#> 5  8 15
#> 6 10 15
#> 7 12 12
#> 
#> $rank
#> $rank$dist
#> [1] 0.5259316
#> 
#> $rank$word
#> [1] 1.5
#> 
#> $rank$qtd
#> [1] 2.807355
#> 
#> $rank$proj
#>       [,1]
#> 3 1.522208

4.All this process can be summarized in the function CSAMiningProcess which performs all the steps listed above.

# CSAMiningProcess
stmotifs <- CSAMiningProcess(D,DS,4,5,4,10,2,2)
rstmotifs[[1]]
#> $isaxcod
#> [1] "bded"
#> 
#> $recmatrix
#>      [,1] [,2] [,3]
#> [1,]    0    0    0
#> [2,]    1    1    1
#> 
#> $vecst
#>    s  t
#> 1  1 11
#> 2  2 11
#> 3  4 17
#> 4  5 17
#> 5  8 15
#> 6 10 15
#> 7 12 12
#> 
#> $rank
#> $rank$dist
#> [1] 0.5259316
#> 
#> $rank$word
#> [1] 1.5
#> 
#> $rank$qtd
#> [1] 2.807355
#> 
#> $rank$proj
#>       [,1]
#> 3 1.522208

2. Visualization

display_motifsDataset(dataset = STMotif::example_dataset, rstmotifs[c(1:4)],  5)

display_motifsSTSeries(dataset = STMotif::example_dataset,rstmotifs[c(1:4)],space = c(1:4,10:12))