The phenomenon of big data has occurred in many fields of knowledge, one of which is astronomy. One example of a large dataset in astronomy is that of numerically integrated time series asteroid orbital elements from a time span of millions to billions of years. For example, the mean motion resonance (MMR) data of an asteroid are used to find out the duration that the asteroid was in a resonance state with a particular planet. For this reason, this research designs a computational model to obtain the mean motion resonance quickly and effectively by modifying and implementing the Symbolic Aggregate Approximation (SAX) algorithm and the motif discovery random projection algorithm on big data platforms (i.e., Apache Hadoop and Apache Spark). There are five following steps on the model: (i) saving data into the Hadoop Distributed File System (HDFS); (ii) importing files to the Resilient Distributed Datasets (RDD); (iii) preprocessing the data; (iv) calculating the motif discovery by executing the User-Defined Function (UDF) program; and (v) gathering the results from the UDF to the HDFS and the .csv file. The results indicated a very significant reduction in computational time between the use of the standalone method and the use of the big data platform. The proposed computational model obtained an average accuracy of 83%, compared with the SwiftVis software.
Keywords: SAX algorithm; big data; mean motion resonance; motif discovery; random projection algorithm; time series.