Main Article Content


Time Series Data mining (TSDM) is one of the most widely used technique that deals with temporal patterns. Genetic algorithm (GA) is a predictive TSDM search technique that is used for solving search/optimization problems. GA is based on the principles and mechanisms of natural selections to find the most nearest optimal solution available from a list of solutions. GA relies on a set of important fundamentals, such as chromosome, crossover and mutation. GA is applied to earthquakes data in the year 2003-2004 in the Suez Gulf in Egypt, gathered from the Egyptian National Seismic Network. The study does not aim to building time series models from the point of time, since the analysis neither include the time nor the prediction of when an earth quake will occur, but to determine the possibility of occurrence of a strong magnitude earthquake after specific sequence of previous earthquakes as temporal pattern. The temporal pattern cluster used is a "circle". The objective function used is a function that gives the highest percentage of correct classification. Empirical results show that crossover and mutation probabilities are 0.4 and .01 respectively for both the training and the testing sample. The algorithm yields 96.98% correct classification for the training sample, and 95.35% for the testing sample.


Temporal Patterns TSDM Genetic Algorithm (GA) Fitness Function Temporal Pattern Cluster.

Article Details

Author Biographies

Sohair F Higazi, Faculty of Commerce Tanta University, Tanta, Egypt

Professor of Applied Statistics

Dept. Applied Statistics and Insurance

Faculty of Commerce, Tanta University, Tanta, Egypt

Walaa Eddien Abdelhadi, Oracle Support Services – EMEA

Senior Fusion Middleware Engineer

Rania M Shalaby, Assistant Lecturer, Dept. of Statistics The Higher Institute for Managerial Science, 6 October

How to Cite
Higazi, S. F., Abdelhadi, W. E., & Shalaby, R. M. (2013). Application of Genetic Algorithm for the Discovery of Hidden Temporal Patterns in Earthquakes Data. Pakistan Journal of Statistics and Operation Research, 9(3), 253-263.