Classification Rules + Time = Temporal Rules

Abstract

Abstract. Due to the wide availability of huge data collection comprising multiple sequences that evolve over time, the process of adapting the classical data-mining techniques, making them capable to work into this new context, becomes today a strong necessity. In [1] we proposed a methodology permitting the application of a classification tree on sequential raw data and the extraction of the rules having a temporal dimension. In this article, we propose a formalism based on temporal first logic-order and we review the main steps of the methodology through this theoretical frame. Finally, we present some solutions for a practical implementation.

Clustering of streaming time series is meaningless

“…Clustering is perhaps the most frequently used data mining algorithm [14], being useful in it's own right as an exploratory technique, and as a subroutine in more complex data mining algorithms [3,5]. Given these two facts, it is hardly surprising that time series clustering has attracted an extraordinary amount of attention [3,7,8,9,11,12,15,16,17,18,20,21,24,25,27,28,29,30,31,32,33,36,38,40,42,45]. The work in this area can be broadly classified into two categories:…”

“…Subsequence clustering is commonly used as a subroutine in many other algorithms, including rule discovery [9,11,15,16,17,20,21,30,32,36,42,45], indexing [27,33], classification [7,8], prediction [37,40], and anomaly detection [45]. For clarity, we will refer to this type of clustering as STS (Subsequence Time Series) clustering.…”

Finding or Not Finding Rules in Time Series

“…Several of the proposed techniques make use of STS clustering (Li et al, 1998;Radhakrishnan et al, 2000). Several techniques for classifying time series make use of STS clustering to preprocess the data before passing to a standard classification technique such as a decision tree Cotofrei & Stoffel, 2002). Clustering of streaming time series has also been proposed as a knowledge discovery tool in its own right.…”

“…We will begin by using the well-known k-means algorithm, since it accounts for the lion's share of all clustering in the time series data mining literature. In addition, the k-means algorithm uses Euclidean distance as its underlying metric, and again the Euclidean distance accounts for the vast majority of all published work in this area Cotofrei & Stoffel, 2002;Das et al, 1998;Fu et al, 2001 Keogh et al, 2001), and as empirically demonstrate in (Keogh & Kasetty, 2002) it performs better than the dozens of other recently suggested time series distance measures.…”

“…As a recent trend, time series data have also been given a lot of attention in the data mining community (Keogh & Kasetty, 2002;Roddick & Spiliopoulou, 2002). This is highly anticipated since time series data has extended Subsequence clustering is commonly used as a subroutine in many other algorithms, including rule discovery (Das et al, 1998;Fu et al, 2001;Uehara & Shimada, 2002;Yairi et al, 2001) indexing (Li et al, 1998;Radhakrishnan et al, 2000), classification Cotofrei & Stoffel, 2002), prediction Tino et al, 2000), and anomaly detection (Yairi et al, 2001). For clarity, we will refer to this type of clustering as STS (Subsequence Time Series) clustering.…”