Informatics Publishing Limited

M. Parimala ¹, K. Poongothai ², S. Sathiyabama ³

Affiliations
1 M. Kumarasamy College of Engineering, Karur, IN
2 Selvam College of Technology, Namakkal, IN
3 K. S. Rangasamy College of Technology, Tiruchengode, IN

Abstract

Mining of periodic patterns in time-series databases is an interesting data mining problem. It can be envisioned as a tool for forecasting and prediction of the future behavior of time-series data. Most researches focused on mining synchronous periodic patterns, but in practice some periodic patterns can not be recognized because of presence of random noisy and disturbance in large datasets. To increase the efficiency of mining Asynchronous periodic patterns on large datasets, our proposal work move in the direction of finding all maximal complex patterns in a single step algorithm using a single dataset scan without mining single event and multiple events patterns explicitly. The asynchronous periodic patterns are mined using depth first search technique. Three parameters are employed to specify the minimum number of repetitions required for a valid segment of non disrupted pattern occurrences, the maximum allowed disturbance between two successive valid segments, and the total repetitions required for a valid sequence.

To find multiple periods based partial periodic patterns, looping over single period using the hit set based approach is one of the naive methods. The algorithm is to directly apply the max-pattern hit set method to each period and the sequence. The proposal evaluates another method for multiple periods using shared mining. This method is similar to the max-pattern single period mining algorithm. During the first scan of the sequence for all periods, the frequent pattern and candidate max-pattern are generated. During the second scan the hit sets of all the periods are generated as in the second scan of max-pattern hit set method. The problem of max-sub pattern tree construction and derivation of frequent patterns from the max-sub pattern tree is also discussed.

The proposal of our work integrates the cyclic partial periodic patterns and incremental patterns with adaptive thresholds to produce interactive mining of partial periodic patterns. The maximum pattern search space and the execution overhead for the proposal work are analyzed with synthetic and real data sets from UCI Repository. In addition it work in the direction of merge mining a generalization of incremental mining which discover patterns of two or more databases that are mined independently of each other. An improved version of merge mining algorithm is planned to build for asynchronous partial periodic patterns in time-series databases.

Keywords

Asynchronous, Cyclic, Incremental Pattern Mining, Periodic Pattern Mining, Sequential.

Full Text

     

K. Poongothai ¹, M. Mathivanan ², T. Senthilkumaran ², S. Revathi ³

Affiliations
1 Selvam College of Technology, Namakkal - 637003, Tamil Nadu, IN
2 A.C.S. College of Engineering, Bengaluru - 560074, Karnataka, IN
3 B. S Abdur Rahman University, Chennai, Tamilnadu, IN

Abstract

Objectives: To deliver excellent performance when compared to existing method Mining Comprehensible Classification Rules for Time-Series (MCCR-TS) using zoo dataset collected from UCI repository comprising of more than 1000 records. Methods/Analysis: Research works on Time-series data has been evolving as a new trend in current scenarios due to the wide range of applications involved in it. One of the widely researched topics is the web usage mining and effective evaluation of time-series movements and transaction associated with them. Most of the studies conducted earlier, focused on identifying the time-series data from the entire logs. However, this type of patterns may not be accurate enough for evaluation due to differentiated behaviors of user patterns are not taken into consideration. Findings: In this paper, we examine the time series for temporal data using integrated cluster-based rule induction mining, by presenting a novel algorithm, namely, Integrated Cluster-based Rule Induction Mining for Time-Series analysis (ICRIM-TS). The algorithm discovers temporal data and the similarities between users are evaluated by the proposed measure, Time Series. To our best knowledge, this is the first work on mining and identification of user behaviors based on the time series data with preferences being given for both user relations and temporal data at the same time. Improvement: Through experimental evaluation, under various settings, the performance of web usage mining is evaluated in terms of precision of users, error threshold value and number of rules extracted.

Keywords

Cluster-based Rule Induction Mining, Temporal Data, Time-series Analysis, User Relations, Web Usage Mining.

Full Text